>> from San Matteo. It's the Cube covering scaler Innovation Day Brought to you by scaler. >> I'm John for the Cube. We are here in San Mateo, California, for special Innovation Day with scaler at their headquarters. Their new headquarters here. I'm here. She here. Lou, Who's Xia Liu? Who's the software engineering team? Good to see you. Thanks for joining. >> Thank you. >> So tell us, what do you do here? What kind of programming? What kind of engineering? >> Sure. Eso i'ma back and suffer engineer at scaler. What I work on from the day to day basis is building our highly scaleable distributed systems and serving our customers fast queries. >> What's the future that you're building? >> Yeah. So one of the project that I'm working on right now is it will help our infrastructure to move towards a more stateless infrastructure s o. The project itself is a meta data storage component and a series of AP ice that Comptel are back and servers where to find a lock file. That might sound really simple, but at the massive scale of ours, it is actually a significant challenge to do it fast and reliably. >> And you're getting date is a big challenge or run knows that data is the new oil date is the goal. Whatever the people saying, the states is super important. You guys have a unique architecture around data ingest What's so unique about it? You mind sharing? >> Of course, s O. We have a lot of things that we do or don't do. Uniquely. I would like to start with the ingestion front of things and what we don't do on that front. So we don't do keywords indexing which most other extinct existing solutions, too. By not doing that, not keeping the index files up to date with every single log message that's incoming. We saved a lot of time and resource, actually, from the moment that our customers applications generate a logline Teo that logline becoming available to for search in scaler. You y that takes just a couple of seconds on DH on other existing solutions. That can take hours. >> So that's the ingests I What about the query side? Because you got in just now. Query. What's that all about? >> Yeah, of course. Actually. Do you mind if we go to black board a little bit? >> Take a look. >> Okay. Grab a chart real quick. Um, so we have a lot of servers around here. We have, uh, Q >> servers. Let's see. >> These are accused servers and, um, a lot of back and servers, Um, just to reiterate on the interest inside a little bit. When locks come in, they will hit one of these Q servers, and you want them Any one of them. And the Q server will kind of batch the log messages together and then pick one of the bag and servers at random and send the batch of locks. Do them any Q can reach any back in servers. And that's how we kind of were able to handle gigs of laughs. How much ever log that you give us way in jazz? Dozens of terabytes of data on a daily basis. Um, and then it is this same farm of back and servers. That's kind of helping us on the query funds crave front. Um, our goal is when a query comes in, we summon all of these back and servers at once. We get all of their computation powers, all of their CPU cores, to serve this one queer Ari, And that is just a massively scalable multi tenant model and in my mind is really economies of scale at its best. >> So scales huge here. So they got the decoupled back in and accused Q system. But yet they're talking to each other. So what's the impact of the customer? What some of the order of magnitude scale we're talking about here? >> Absolutely. So for on the loch side, we talked about seconds response time from logs being generated, too. They see the lock show up and on the query side, um, the median response time of our queries is under 100 milli second. And we defined that response time from the moment the customer hit in the return button on their laptop to they see results show up and more than 90% of our queries return results in under one second. >> So what's the deployment model for the customers? So I'm a customer. Oh, that sounds great. Leighton sees a huge issue one of low late and seek. His legacy is really the lag issue for data. Do I buy it as a service on my deploying boxes? What does this look like here? >> Nope. Absolutely. Adult were 100 plan cloud native. All of this is actually in our cloud infrastructure and us a customer. You just start using us as a sulfur is a service, and when you submit a query, all of our back and servers are at your service. And what's best about this model is that asks Keller's business girls. We will add more back and servers at more computation power and you as a customer's still get all of that, and you don't need to pay us any extra for the increased queries. >> What's the customer news case for this given you, given example of who would benefit from this? >> Absolutely. So imagine your e commerce platform and you're having this huge black Friday sales. Seconds of time might mean millions of revenues to you, And you don't wantto waste any time on the logging front to debug into your system to look at your monitoring and see where the problem is. If you ever have a problem, so we give you a query response time on the magnitude of seconds versus other is existing solutions. Maybe you need to wait for minutes anxiously in front of your computer. >> She What's the unique thing here? This looks like a really good actor, decoupling things that might make sense. But what's the What's the secret sauce? You? What's the big magic here? >> Yeah, absolutely. So anyone can kind of do a huge server farm Route Fours query approach. But the 1st 80% of a brute force algorithm is easy. It's really the last 20%. That's kind of more difficult, challenging and really differentiate. That's from the rest of others. Solutions. So to start with, we make every effort we can teo identify and skip the work that we don't have to do. S O. Maybe we can come back to your seats. >> Cut. >> Okay, so it's so it's exciting. >> Yeah. So we there are a couple things we do here to skip the work that we don't have to do. As we always say, the fastest queries are those we don't even have to run, which is very true. We have this Colin, our database that wee boat in house highly performance for our use case that can lead us only scan the columns that the customer cares about and skipped all the rest. And we also build a data structure called bloom Filters And if a query term does not occur in those boom filters, we can just skip the whole data set that represents >> so that speed helps on the speed performance. >> Absolutely. Absolutely. If we don't even have to look at that data set, >> You know, I love talking to suffer engineers, people on the cutting edge because, you know, you guys were startup. Attracting talent is a big thing, and people love to work on hard problems. What's the hard problem that you guys are solving here? >> Yeah, absolutely. S o we we have this huge server farm at at our disposal. It's, however, as we always say, the key to brute force algorithms is really to recruit as much force as possible as fast as we can. If you have hundreds thousands, of course lying around. But you don't have an effective way to some of them around when you need them. Then there's no help having them around 11 of the most interesting things that my team does is we developed this customised scatter gather algorithm in order to assign the work in a way that faster back and servers will dynamically compensate for slower servers without any prior knowledge. And I just love that >> how fast is going to get? >> Well, I have no doubt that will one day reach light speed. >> Specialist. Physics is a good thing, but it's also a bottleneck. Just what? Your story. How did you get into this? >> Yeah, s o. I joined Scaler about eight months ago as an ap s server, Actually. Sorry. As an FBI engineer, actually eso during my FBI days. I use scaler, the product very heavily. And it just became increasingly fascinated about the speed at which our queria runs. And I was like, I really want to get behind the scene and see what's going on in the back end. That gives us such fast query. So here I am. Two months ago, I switched the back and team. >> Well, congratulations. And thanks for sharing that insight. >> Thank you, John. Thank >> jumper here with Cuban Sites Day and Innovation Day here in San Mateo. Thanks for watching

>> Live from Orlando, Florida. It's theCUBE. Covering Microsoft Ignite. Brought to you by Cohesity, and theCUBE's ecosystem partners. >> Welcome back to the CUBE's live coverage of Microsoft Ignite here in Orlando, Florida. I'm your host, Rebecca Knight. Along with my co-host Stu Miniman. We're joined by Andrew Liu. He is the senior product manager at Azure Cosmos DB. Thanks so much for coming on the show Andrew. >> Oh, thank you for hosting. >> You're a first timer, so this will be a lot of fun. So, talk to me a little bit. Azure Cosmos DB is a database for building blazing fast planet scale applications. Can you tell our viewers a little bit about what you do and about the history of Azure Cosmos? >> Sure, so Azure Cosmos DB started with, about eight years ago, where we were also outgrowing a lot of our own database needs with what we had previously built. And a lot of the challenges that we had was really around partitioning, replication, and resource governance. So, I'll talk a little bit about each one. Partitioning is really about solving the problem of scale. Right? I have so much data, doesn't fit on a single machine, and I have so many requests per second. Also doesn't, can't be served out of a single machine. So how do I go and build a system, a database that can elastically scale over a cluster of machines, so I don't have to manually shard, and as a user have to shard a database across many, many instances. This way I really want to be able to scale just seamlessly. The velocity problem is, we also wanted to build something that, can respond in a very fast manner, in terms of latency. So, it's great and all that we can serve lots of request per second, but, what is the response time of each one of those requests? And the resource governance was there to really actually build this as a cloud native database in which we wanted to exploit the properties of our cloud. We wanted to use the economies of scale that we can have basically data centers built all around the world, and build this as a multi, truly multi-tenant service. And by doing so we can also afford the total cost of ownership for us, as well as, a guaranteed predictable performance for the tenants. Now we did this, for initially our first party tenants at Microsoft, where we have made a bet on everything from our Microsoft live platform, to Office, to Azure itself as built on Azure Cosmos DB. And about four years ago we found that hey, this is not really just a Microsoft problem that we're solving, but it's an everybody problem, it's become universal, and so we've launched it out to the open. >> Yeah, Andrew that's, great point, and I want you to help unpack that for us a little bit because you know, we've been saying on theCUBE for many years, distributed architectures are some of the toughest challenges of our time, but, if I'm a Facebook, or a Google, or a Microsoft, I understand some of the challenges, and I understand why I need it, but, when you talk about scale, well, scale means a lot of different things to a lot of different people. So, how does Cosmos? What does that mean to your users, end users, why do they need this? You know, haven't they just felt some microservices architecture? And they'll just leverage, ya know what's in Azure. And things like that. How does this global scale impact the typical user? >> So I'm actually seeing this come in different types of patterns for different types of industries. So for example, in manufacturing we're commonly seeing Cosmos DB used really for that scalability for the write scalability, and having many, many concurrent writes per second. Typically this is done in an IoT telemetry, or an IoT device registry case. So let's use one of our customers for example, Toyota. Each year they're shipping millions of vehicles on the road, and they're building a big connected car platform. The connected car platform allows you to do things like, whenever it alerts an airbag gets deployed, they can go and make sure and call their driver, hey, I saw the airbag was deployed are you okay? And if the user doesn't pick up their phone, immediately notify emergency services. But the challenge here is if each year I'm shipping millions of vehicles on the road, and each of 'em has a heartbeat every second, I'm dealing with millions of writes per second, and I need a database that can scale to that. In contrast, in retail I'm actually seeing very different use cases. They're using more of the replication side of our stock where they have a global user base, and they're trying to expand an eCommerce shop. So for example ASOS is a big fashion retailer, they ship to 200 different countries globally, and they want to make sure that they can deliver real-time experiences like real-time personalization, and based off of who the user is recommended set of products that is tailored to that user. Well now what I need is a data set that can expand to my shoppers across two different hundred, 200 countries around the globe, and deliver that with very, very low latency so that my web experience is also very robust. So what they use is our global distribution, and our multi-mastering technology. Where we can actually have a database presence, similar to like what a CDN does for static content, we're doing for our dynamic evolving content. So in a database your work load, typically your data set is evolving, and you want to be able to run queries with consistency over that. As opposed to in CDN you're typically serving static assets. Well here we can actually support those dynamic content, and then build these low latency experiences to users all around the globe. The other area we see a lot of usage is in ISV's for mission critical workloads. And the replication actually gets us two awesome properties, right? One is the low latency by shipping data closer to where the user is, but the other property you get is a lot of redundancy, and so we actually also offer industry leading SLA's where we guarantee five nines of availability, and the way we're able to do so is, with a highly redundant architecture you don't care if let's say a machine were to bomb out at any given time, because we have multiple redundant copies in different parts of the globe. You're guaranteed that your workload is always online. >> So my question for you is, when you have these, you just described some really, really interesting customer use cases in manufacturing, in retail, do you then create products and services for each of these industries? Or do you say hey other retail customers, we've noticed this really works for this customer over here, how do you go out to the community with what you're selling? >> Ah, got it. So we actually have found that this can be a challenging space for some of our customers today, 'cause we have so many products. The way we kind of view it is we want to have a portfolio, so that you can always choose the right tool for the right job. And I think a lot of how Microsoft has evolved as a business actually is around this. Previously we would sell a hammer, and we'd tell you don't worry everything's a nail, even if it looks like a screw let's just pretend it's a nail and whack it down. But today we've built this big vast toolbox, and you can think of Cosmos DB as just one of many tools in our vast toolbox. So if you have a screw maybe you pickup a screwdriver, and screw that in. And the way Azure works is then if we have a very comprehensive toolbox, depending on what precise scenario you have, you can kind of mix and match the tools that fit your problem. So think of them as like individual Lego blocks, and whether you're building like a death star, or an x-wing, you can go, and assemble the right pieces for your application. >> Andrew, some news at the show around Cosmos DB. Share us what the updates are. >> Oh sure, so we're really excited to launch a few new features. The highlights are multi-master, and Cassandra API. So multi-master really exploits the replicated nature of our database. Before multi-master what we would do is, we would allow you to have a globally distributed database in which you can have write requests go to single region, and reads being served out of any of these other locations. With multi-master we've actually made it so that each of those replicas we've deployed around the globe can also accept write requests. What that translates to from a user point of view is number one, your write requests are a lot faster, they're super low latency, single-digit millisecond latency in fact. No matter where the user is around the globe. And number two, you also get much higher write availability. So even if let's say, we're having a natural disaster, we had a nasty hurricane as you know pass through on the east coast last week, but with a globally distributed database the nice thing is even if you have, let's say, a power disruption in one region of the world, it doesn't matter cause you can then just fail over, and talk to another data center, where you have a live replica already located. So we just came out with multi-master. The short summary is low latency writes, as well as high available writes. The other feature that we launched is Cassandra API, and as you know this is a multi-model, multi-API database. What that means is, what we're trying to do is also meet our users where they are. As opposed to pushing our proprietary software on them, and we take the whole concept of vendor lock-in very, very seriously. Which is why we make such a big bet on the open source ecosystem. If you already have, let's say a MongoDB application, or a Cassandra application, but you'd really love to be able to take advantage of some of the novel properties that we've built with building a fully managed multi-master database. Well, what we've done is we've implemented this as a wire level protocol on the server side. So it can take an existing application, not change a single line of code, and point it to Cosmos DB as a back-end, and then take advantage of Cosmos DB as your database. >> One of the interesting things if you look at the kind of changing face of databases, it's how users are being able to leverage their data. You talk about everything from you know, I think Cassandra back, and some of the big data discussions, today everything's AI which I know is near and dear to Microsoft's heart. Satya Nadella I'm talking about, how do you think of the role of data in this solution set? >> Sorry, can you say that one more time? >> So, how customers think about leveraging data, how things like Cosmos allow them to really extract the value out of data, not just be some database that kind of stuck in the back-end somewhere. >> Yeah, yeah. I mean a lot of it is the new novel experiences people are building. So for example, like the connected car platform, I'm seeing people actually build this, and take advantage of new novel territories that a traditional automobile manufacturer used to not do. Not only are they building experiences around, how do they provide value to their end users? Like the air bag scenario, but they're also using this as a way of building value for their business, and how to make sure that, hey when, next time you're up for an oil change that they can send a helpful reminder, and say hey I noticed you're due for an oil change in terms of mileage. Why don't I just go set up an appointment, just up for you, as well as other experiences for things, like when they want to do fleet management, and do partnerships with either ride sharing companies like Uber, and Lyft, or rental car companies like Avis, Hertz, et cetera. I've also seen people take advantage of, taking kind of new novel experiences through databases, through AI, and machine learning. So for example, the product recommendations. This was something that historically, when I wanted to do recommendations a decade ago, maybe I have some big beefy data lake running somewhere in the back-end, it might take a week to munch through that data, but that's okay, a week later once I'm ready, I'll send out some mail, maybe some email to you, but today when I want to actually show live right when the user is browsing my website, my website has to load fast right? If my goal is to increase conversions on sales, having a slow running website is the fastest way for my user to click the back button. But if I want to build real-time personalization, and want to generate let's say a recommendation within 200 millisecond latency, well now that I have databases that can guarantee me single-digit millisecond latency, it gives me ample time to actually improve the business logic for those recommendations. >> I want to ask you a question about culture, because you are based at the mothership in Redmond, Washington. So we heard Satya Nadella on the main stage today talk about tech intensiveness, tech intensity, sorry, this idea that we need to not only be adopting technology, but also building the latest, and greatest. I'm curious about, how that translates at Microsoft's campus, and sort of how, how this idea is, infuses how you work with your colleagues, and then also how you work with your customers and partners? >> I think some of the biggest positive changes I've seen over the last decade has been how much more of a customer focus we have today then ever. And i think a lot of things have led to that. One is, just the ability to ship much faster. As we move to Cloud services we're no longer in these big box product release cycles of building a product, and waiting like one or two years to ship it to our users. But now we can actually get some real-time feedback. So as we go, and ship, and deploy software, we actually deploy even on a weekly cadence over here. What that allows us to do is actually experiment a lot more, and get real-time feedback, so if we have an idea, and rather than having to go through a long lengthy vetting process, spending years building, and hoping that it really pays off. What we can do is we can just go talk to our users, and say hey, ya know, we have an idea for our future. We'd love to get your feedback, or a lot of times honestly our customers actually come to us, where we're so tightly engaged these days, that when, users even come to us, and say like hey, what do you think about this idea? It would really add a lot of value to my scenario. We go, and try to root cause that, really get an idea of what exactly that they need. But then we can turn that around in blazing fast time. And I think a lot of the shift to Cloud services, and being able to avoid the overhead of well we got to wait for this ship train, and then wait for the right operation personnel to go and deploy the updates. Now that we can control our own destiny, and just ship on a very, very fast cadence, we're closer to our users, and we experiment a lot more, and I think it's a beautiful thing. >> Great, well Andrew thank you so much for coming on theCUBE, it was fun talking to you. >> Oh yeah, thank you for hosting. >> I'm Rebecca Knight, for Stu Miniman, we will have more from theCUBE's live coverage of Microsoft Ignite coming up just after this. (techno music)

>> Announcer: Live from New York City It's TheCube. Covering CyberConnect 2017. Brought to you by Centrify and the Institute for Critical Infrastructure Technology. >> It got out of control, they were testing it. Okay, welcome back everyone. We are here live in New York City for CyberConnect 2017. This is Cube's coverage is presented by Centrify. It's an industry event, bringing all the leaders of industry and government together around all the great opportunities to solve the crisis of our generation. That's cyber security. We have Cricket Liu. Chief DNS architect and senior fellow at Infoblox. Cricket, great to see you again. Welcome to theCUBE. >> Thank you, nice to be back John. >> So we're live here and really this is the first inaugural event of CyberConnect. Bringing government and industry together. We saw the retired general on stage talking about some of the history, but also the fluid nature. We saw Jim from Aetna, talking about how unconventional tactics and talking about domains and how he was handling email. That's a DNS problem. >> Yeah, yeah. >> You're the DNS guru. DNS has become a role in this. What's going on here around DNS? Why is it important to CyberConnect? >> Well, I'll be talking tomorrow about the first anniversary, well, a little bit later than the first anniversary of the big DDoS attack on Dyn. The DNS hosting provider up in Manchester, New Hampshire. And trying to determine if we've actually learned anything, have we improved our DNS infrastructure in any way in the ensuing year plus? Are we doing anything from the standards, standpoint on protecting DNS infrastructure. Those sorts of things. >> And certainly one of the highlight examples was mobile users are masked by the DNS on, say, email for example. Jim was pointing that out. I got to ask you, because we heard things like sink-holing addresses, hackers create domain names in the first 48 hours to launch attacks. So there's all kinds of tactical things that are being involved with, lets say, domain names for instance. >> Cricket: Yeah, yeah. >> That's part of the critical infrastructure. So, the question is how, in DDoS attacks, denial-of-service attacks, are coming in in the tens of thousands per day? >> Yeah, well that issue that you talked about, in particular the idea that the bad guys register brand new domain names, domain names that initially have no negative reputation associated with them, my friend Paul Vixie and his new company Farsight Security have been working on that. They have what is called a -- >> John: What's the name of the company again? >> Farsight Security. >> Farsight? >> And they have what's called a Passive DNS Database. Which is a database basically of DNS telemetry that is accumulated from big recursive DNS servers around the internet. So they know when a brand new domain name pops up, somewhere on the internet because someone has to resolve it. And they pump all of these brand new domain names into what's called a response policy zone feed. And you can get for example different thresh holds. I want to see the brand new domain names created over the last 30 minutes or seen over the last 30 minutes. And if you block resolution of those brand new domain names, it turns out you block a tremendous amount of really malicious activity. And then after say, 30 minutes if it's a legitimate domain name it falls off the list and you can resolve it. >> So this says your doing DNS signaling as a service for new name registrations because the demand is for software APIs to say "Hey, I want to create some policy around some techniques to sink-hole domain address hacks. Something like that? >> Yeah, basically this goes hand in hand with this new system response policy zone which allows you to implement DNS policy. Something that we've really never before done with DNS servers, which that's actually not quite true. There have been proprietary solutions for it. But response policy zones are an open solution that give you the ability to say "Hey I do want to allow resolution of this domain name, but not this other domain name". And then you can say "Alright, all these brand new domain names, for the first 30 minutes of their existence I don't want-- >> It's like a background check for domain names. >> Yeah, or like a wait list. Okay, you don't get resolved for the first 30 minutes, that gives the sort of traditional, reputational, analyzers, Spamhaus and Serval and people like that a chance to look you over and say "yeah, it's malicious or it's not malicious". >> So serves to be run my Paul Vixie who is the contributor to the DNS protocol-- >> Right, enormous contributor. >> So we should keep an eye on that. Check it out, Paul Vixie. Alright, so DNS's critical infrastructure that we've been talking about, that you and I, love to riff about DNS and the role What's it enabled? Obviously it's ASCII, but I got to ask you, all these Unicode stuff about the emoji and the open source, really it highlight's the Unicode phenomenon. So this is a hacker potential haven. DNS and Unicode distinction. >> It's really interesting from a DNS standpoint, because we went to a lot of effort within the IETF, the Internet Engineering Task Force, some years ago, back when I was more involved in the IETF, some people spent a tremendous amount of effort coming up with a way to use allow people to use Unicode within domain name. So that you could type something into your browser that was in traditional or simplified Chinese or that was in Arabic or was in Hebrew or any number of other scripts. And you could type that in and it would be translated into something that we call puny code, in the DNS community, which is an ASCII equivalent to that. The issue with that though, becomes that there are, we would say glifs, most people I guess would say characters, but there are characters in Unicode that look just like, say Latin alphabet characters. So there's a lowercase 'a' for example, in cyrillic, it's not a lowercase 'a' in the Latin alphabet, it's a cyrillic 'a', but it looks just like an 'a'. So it's possible for people to register names, domain names, that in there Unicode representation, look like for example, PayPal, which of course has two a's in it, and those two a's could be cyrillic a's. >> Not truly the ASCII representation of PayPal which we resolve through the DNS. >> Exactly, so imagine how subtle an attack that would be if you were able to send out a bunch of email, including the links that said www.-- >> Someone's hacked your PayPal account, click here. >> Yeah, exactly. And if you eyeballed it you'd think Well, sure that's www.PayPal.com, but little do you know it's actually not the -- >> So Jim Ruth talked about applying some unconventional methods, because the bad guys don't subscribe to the conventional methods . They don't buy into it. He said that they change up their standards, is what I wrote down, but that was maybe their sort of security footprint. 1.5 times a day, how does that apply to your DNS world, how do you even do that? >> Well, we're beginning to do more and more with analytics DNS. The passive DNS database that I talked about. More and more big security players, including Infoblox are collecting passive DNS data. And you can run interesting analytics on that passive DNS data. And you can, in some cases, automatically detect suspicious or malicious behavior. For example you can say "Hey, look this named IP address mapping is changing really, really rapidly" and that might be an indication of let's say, fast flux. Or you can say "These domain names have really high entropy. We did an engram analysis of the labels of these". The consequence of that we believe that this resolution of these domain names, is actually being used to tunnel data out of an organization or into an organization. So there's some things you can do with these analytical algorithms in order to suss out suspicious and malicious. >> And you're doing that in as close to real time as possible, presumably right? >> Cricket: That's right. >> And so, now everybody's talking about Edge, Edge computing, Edge analytics. How will the Edge effect your ability to keep up? >> Well, the challenge I think with doing analytics on passive DNS is that you have to be able to collect that data from a lot of places. The more places that you have, the more sensors that you have collecting passive DNS data the better. You need to be able to get it out from the Edge. From those local recursive DNS servers that are actually responding to the query's that come from say your smart phone or your laptop or what have you. If you don't have that kind of data, you've only got, say, big ISPs, then you may not detect the compromise of somebody's corporate network, for example. >> I was looking at some stats when I asked the IOT questions, 'cause you're kind of teasing out kind of the edge of the network and with mobile and wearables as the general was pointing out, is that it's going to create more service area, but I just also saw a story, I don't know if it's from Google or wherever, but 80% plus roughly, websites are going to have SSL HTBS that they're resolving through. And there's reports out here that a lot of the anti virus provisions have been failing because of compromised certificates. And to quote someone from Research Park, and we want to get your reaction to this "Our results show", this is from University of Maryland College Park. "Our results show that compromised certificates pose a bigger threat than we previously believed, and is not restricted to advanced threats and digitally signed malware was common in the wild." Well before Stuxnet. >> Yeah, yeah. >> And so breaches have been caused by compromising certificates of actual authority. So this brings up the whole SSL was supposed to be solving this, that's just one problem. Now you've got the certificates, well before Stuxnet. So Stuxnet really was kind of going on before Stuxnet. Now you've got the edge of the network. Who has the DNS control for these devices? Is it kind of like failing? Is it crumbling? How do we get that trust back? >> That's a good question. One of the issues that we've had is that at various points, CAs, Certificate Authorities, have been conned into issuing certificates for websites that they shouldn't have. For example, "Hey, generate a cert for me". >> John: The Chinese do it all the time. >> Exactly. I run www. Bank of America .com. They give it to the wrong guy. He installs it. We have I think, something like 1,500 top level certification authorities. Something crazy like that. Dan Komenski had a number in one of his blog posts and it was absolutely ridiculous. The number of different CA's that we trust that are built into the most common browsers, like Chrome and Firefox and things like that. We're actually trying to address some of those issues with DNS, so there are two new resource records being introduced to DNS. One is TLSA. >> John: TLSA? >> Yeah, TLSA. And the other one is called CAA I think, which always makes me think of a California Automotive Association. (laughter) But TLSA is basically a way of publishing data in your own zone that says My cert looks like this. You can say "This is my cert." You can just completely go around the CA. And you can say "This is my cert" and then your DNS sec sign your zone and you're done. Or you can do something short of that and you can say "My cert should look like this "and it should have this CA. "This is my CA. "Don't trust any other one" >> So it's metadata about the cert or the cert itself. >> Exactly, so that way if somebody manages to go get a cert for your website, but they get that cert from some untrustworthy CA. I don't know who that would be. >> John: Or a comprimised-- >> Right, or a compromised CA. No body would trust it. No body who actually looks up the TSLA record because they'll go "Oh, Okay. I can see that Infoblox's cert that their CA is Symantech. And this is not a Symantech signed cert. So I'm not going to believe it". And at the same time this CAA record is designed to be consumed by the CA's themselves, and it's a way of saying, say Infoblox can say "We are a customer of Symantech or whoever" And when somebody goes to the cert and says "Hey, I want to generate a certificate for www.Infoblox.com, they'll look it up and say "Oh, they're a Symantech customer, I'm not going to do that for you". >> So it creates trust. So how does this impact the edge of the network, because the question really is, the question that's on everyone's mind is, does the internet of things create more trust or does it create more vulnerabilities? Everyone knows it's a surface area, but still there are technical solutions when you're talking about, how does this play out in your mind? How does Infoblox see it? How do you see it? What's Paul Vixie working on, does that tie into it? Because out in the hinterlands and the edge of the network and the wild, is it like a DNS server on the device. It could be a sensor? How are they resolving things? What is the protocol for these? >> At least this gives you a greater assurance if you're using TLS to encrypt communication between a client and a web server or some other resource out there on the internet. It at least gives you a better assurance that you really aren't being spoofed. That you're going to the right place. That your communications are secure. So that's all really good. IOT, I think of as slightly orthogonal to that. IOT is still a real challenge. I mean there is so many IOT devices out there. I look at IOT though, and I'll talk about this tomorrow, and actually I've got a live event on Thursday, where I'll talk about it some more with my friend Matt Larson. >> John: Is that going to be here in New York? >> Actually we're going to be broadcasting out of Washington, D.C. >> John: Were you streaming that? >> It is streamed. In fact it's only streamed. >> John: Put a plug in for the URL. >> If you go to www.Infoblox.com I think it's one of the first things that will slide into your view. >> So you're putting it onto your company site. Infoblox.com. You and Matt Larson. Okay, cool. Thursday event, check it out. >> It is somewhat embarrassingly called Cricket Liu Live. >> You're a celebrity. >> It's also Matt Larson Live. >> Both of you guys know what you're talking about. It's great. >> So there's a discussion among certain boards of directors that says, "Look, we're losing the battle, "we're losing the war. "We got to shift more on response "and at least cover our butts. "And get some of our response mechanisms in place." What do you advise those boards? What's the right balance between sort of defense perimeter, core infrastructure, and response. >> Well, I would certainly advocate as a DNS guy, that people instrument their DNS infrastructure to the extent that they can to be able to detect evidence of compromise. And that's a relatively straight forward thing to do. And most organizations haven't gone through the trouble to plumb their DNS infrastructure into their, for example, their sim infrastructure, so they can get query log information, they can use RPZs to flag when a client looks up the domain name of a known command and control server, which is a clear indication of compromise. Those sorts of things. I think that's really important. It's a pretty easy win. I do think at this point that we have to resign ourselves to the idea that we have devices on our network that are infected. That game is lost. There's no more crunchy outer shell security. It just doesn't really work. So you have to have defensive depth as they say. >> Now servs has been around for such a long time. It's been one of those threats that just keeps coming. It's like waves and waves. So it looks like there's some things happening, that's cool. So I got to ask you, CyberConnect is the first real inaugural event that brings industry and some obviously government and tech geeks together, but it's not black hat or ETF. It's not those geeky forums. It's really a business community coming together. What's your take of this event? What's your observations? What are you seeing here? >> Well, I'm really excited to actually get the opportunity to talk to people who are chiefly security people. I think that's kind of a novelty for me, because most of the time I think I speak to people who are chiefly networking people and in particular that little niche of networking people who are interested in DNS. Although truth be told, maybe they're not really interested in DNS, maybe they just put up with me. >> Well the community is really strong. The DNS community has always been organically grown and reliable. >> But I love the idea of talking about DNS security to a security audience. And hopefully some of the folks we get to talk to here, will come away from it thinking oh, wow, so I didn't even realize that my DNS infrastructure could actually be a security tool for me. Could actually be helpful in any way in detecting compromise. >> And what about this final question, 'cause I know we got a time check here. But, operational impact of some of these DNS changes that are coming down from Paul Vixie, you and Matt Larson doing some things together, What's the impact of the customer and they say "okay, DNS will play a role in how I role out my architecture. New solutions for cyber, IOT is right around the corner. What's the impact to them in your mind operationally. >> There certainly is some operational impact, for example if you want to subscribe to RPZ feeds, you've got to become a customer of somebody who provides a commercial RPZ feed or somebody who provides a free RPZ feed. You have to plumb that into your DNS infrastructure. You have to make sure that it continues transferring. You have to plumb that into your sim, so when you get a hit against an RPZ, you're notified about it, your security folks. All that stuff is routine day to day stuff. Nothing out of the ordinary. >> No radical plumbing changes. >> Right, but I think one of the big challenges in so many of the organizations that I go to visit, the security organization and the networking organization are in different silos and they don't necessarily communicate a lot. So maybe the more difficult operational challenge is just making sure that you have that communication. And that the security guys know the DNS guys, the networking guys, and vice versa. And they cooperate to work on problems. >> This seems to be the big collaboration thing that's happening here. That it's more of a community model coming together, rather than security. Cricket Liu here, DNS, Chief Architect of DNS and senior fellow of Infoblox. The legend in the DNS community. Paul Vixie amongst the peers. Really that community holding down the fort I'll see a lot of exploits that they have to watch out for. Thanks for your commentary here at the CyberConnect 2017 inaugural event. This is theCUBE. We'll be right back with more after this short break. (techno music)

>> Hello, we are here On the Ground. This is theCUBE's On the Ground program at Centrify's Headquarters. We go to Cricket Liu, chief DNS officer at Infoblox. Been with the company from the beginning. Great to see you again. Wrote the book on DNS. What year was that? That was between DNS, was like, when I was born. >> Yeah, 1992. September 1992 was when it was published. >> Great to see you. We've done some podcasts together over the years. >> Yeah, good to see you too. >> DNS, now obviously global, ICANN's now global, it's part of the U.N., all different governance bodies, but it's certainly still critical infrastructure. >> Yeah, absolutely. >> Critical infrastructure is now the big conversation as the security paradigm has moved from data center to the Cloud, there's no perimeter anymore. >> Yeah. >> How is that changing the DNS game? >> Well, I think that folks are starting to realize how critical DNS is. In October of last year, we had that huge DDoS attack against Dyn, the big DNS hosting provider in New Hampshire and I think that woke a lot of folks up. A lot of folks realized, holy cow, these guys are not too big to fail as they say. Even though they have enormous infrastructure, widely distributed around the globe, they have such a concentrational power that a huge number of really, really popular web properties were inaccessible for quite sometime, so I think that caused a lot of people to look at their own DNS infrastructure and to reevaluate it and say, well maybe I need to do something. >> Interesting about the stack wars that are going on, that attack, as we've lived through and you've been part of it as chief technical officer in many companies. DNS was always that part where it'd be secure but now you have block change, you have new kinds of infrastructure with mobile computing now over 10 years post iPhone. >> Yep, the critical moment. >> How has infrastructure changed, beyond DNS 'cause it still needs to work together? >> Yeah, well, it's funny because we do have all of these new types of devices. We do have new technologies. But a lot of things have remained the same. DNS is still the same. The remarkable thing is that the latest version in my book is 10 years old, actually 11 years old now, so it's older than the iPhone and people still buy it because the underlying theory is still the same. It hasn't changed. It's a testament, really, to the quality of the original design of DNS that it still works for anything and that it's scaled to serve a network as diverse and as large as the internet is today. >> What's your biggest observation, looking back over the past decade with DNS, about the emergence of virtual machines, now Cloud. Again, the game is still the same 'cause DNS is the plumbing and it provides a lot of the key critical infrastructure for the web and now mobile. What's the biggest observations that you've seen over the decade? >> Well I'd say one of the things that's happened over the last several years that's maybe the most important development in DNS is something that we call response policy zones. Up until now, DNS servers have just been sort of blithely complicit when it comes to, for example, malware. Malware wakes up on a device and it assumes that it has DNS available to it and it uses DNS, for example, to find command to control server, maybe a drop server to exfiltrate data to. In the DNS server, even though it's being asked to look up the address record for CommandAndControlServer.Malware.Org, it just happily goes along with it. A few years ago, Paul Vixie, who I've known for a very long time, came up with this idea called response policy zones which is basically to imbue our DNS servers with resolution policy so that you can tell them, hey if you get a query for a domain name that we know is being used maliciously, don't answer it. Don't resolve it like you normally do. Instead, hand back a little white lie like that doesn't exist and moreover, log the fact that somebody looked it up because it's a good indication that they're infected. >> So bringing policy to DNS is really making it more intelligent. >> Yeah, that's right. >> And certainly as networks grow, I was just watching some of my friends setting up the wireless at Burning Man and the whole new change of how Wi-Fi is being deployed and how networks are being constructed is really coming down to some of the basic principles of DNS to route more, be responsive, and this is kind of a new change. >> Yeah, there's a lot going on in changes to the deployment of DNS. It used to be that most big companies ran all their own DNS infrastructure. At this point, I think most large companies don't bother running, for example, what we'd call their external authoritative DNS infrastructure. They give that to a big hosting provider to do, somebody like Dyn or Verisign or Neustar or somebody like that, so that's a big change. >> Cricket, I want to ask you about the CyberConnect Event going on in New York. Infoblox is involved. Security is paramount, so now an industry event. Centrify is the main sponsor. You guys are involved as a vendor, but it's not a vendor event, it's a industry event. It's a broad category. What's your thoughts on this kind of industry event? Usually in events it's been Black Hat or vendor events pushing their wares and selling their stuff but now security is global. What's your take on this event? >> Well, I'm hoping to be able to spend a little bit of time talking to folks who come to the event about DNS and how it can be used as a tool in their security tool chain. The folks who come to us as Infoblox to our events already know about DNS. They're already network administrators or they're responsible for DNS or something like that. My hope is that we can reach a broader audience through CyberConnect and actually talk to folks who maybe haven't considered DNS as a security tool. Who maybe haven't thought about the necessity to bolster their DNS infrastructure. >> One final question since we're on bonus material time. I've got to ask you about the global landscape. I mean, in my early days involved in DNS when I came was from the '98 to the 2000 time frame. International domain names were Unicode. That's not ASCII. So that technically wasn't DNS, but still, they were keywords. They had this global landscape in, say, China, that actually wasn't DNS so there's all these abstraction layers. Has anything actually evolved out of that trend of really bringing an abstraction layer on top of DNS and certainly now with the nation-states with security are issues, China, Russia, et cetera. How does all that play out? >> Well, international domain names have actually taken off in some areas. And basically it's as you say, you have the ability now to use Unicode labels in domain names in certain contexts, for example, if you're using your web browser you can type in a Unicode domain name and then what the web browser does is it translates it into an equivalent ASCII representation and then resolves it using DNS which is the traditional DNS that doesn't actually know about Unicode. There are actually some very interesting security implications to using Unicode. For example, people can register things that have Unicode, we would say, glyphs in them that look exactly like regular ASCII characters. For example, you could register paypal.com where the A's are actually lowercase A's in Cyrillic. It's not the same code point as an ASCII A. So it's visually. >> Great for hackers. >> Oh yeah. Visually indistinguishable from paypal.com in a lot of contexts and people might click on it and go to a page that looks like PayPal's. >> John: So its a phishing dream. >> Yeah, really dangerous potentially and so we're working out some of the implications of that, trying to figure out, within, for example, web browsers, how do we protect the user from things like this? >> And a lot of SSL out there, now you're seeing HTTPS everywhere. Is that now the norm? >> Yeah, actually, within the internet engineering task force, the IETF, after it became obvious that state-sponsored-- >> John: Attacks. >> Eavesdropping. >> You were smiling. >> Was kind of the norm. >> Got to find the right word. >> Yeah, the IETF embarked on an effort called DPRIVE and DPRIVE is basically a bunch of individual tracks to encrypt basically every single part of the DNS channel, especially that between what we call a stub resolver and the recursive DNS server so that if you're a customer here in the United States and a subscriber to an ISP like Comcast or whomever, you can make sure that that first hop between your computer and the ISP is secured. >> We're getting down and dirty under the hood with Cricket Liu on DNS. I got to ask kind of up level to the consumer. One of the things that kind of pisses me off the most when I'm surfing the web is you see the browser doesn't resolve or you go hit someone's website, oh yeah, something.io, these new domain names, top level gTLDs are out there, .media, all these, and companies have firewalls or whatever their equipment is and it doesn't let it through. Because they're trying to protect the perimeter still, must be, I mean, what does that mean when companies aren't letting those URLs then, it is a firewall issue or is it more they're still perimeter based, they're not resolving it, they're afraid of malware? Somethings aren't resolving in? What does that mean? >> Well I think as often as not it's an operational problem. It could be just a misconfiguration on the part of the folks who are hosting the target website's DNS. It could be that. I don't know a lot of folks who-- >> So it's one of their policies or something, it's just kind of locking down. >> Could be that too. Or it could be, for example, that they have a proxy server and they're trying to limit access to the internet by category. Maybe it does categorization and filtering by-- >> Can you work on that? Can you write some code for that? Well thanks, great to see you, thanks for sharing this conversation here On The Ground at Centrify. >> You're welcome. >> And good luck with the CyberConnect Conference. >> Yeah, nice to see you too. >> Alright, I'm John Furrier with On The Ground here on theCUBE at Centfity's headquarters in Silicon Valley. Thanks for watching.

>> Welcome back everyone to the closing remarks here before we kick off our ecosystem portion of the program. We're live in Palo Alto for theCUBE special presentation of Supercloud 2. It's the second edition, the first one was in August. I'm John Furrier with Dave Vellante. Here to wrap up with our special guest analyst George Gilbert, investor and industry legend former colleague of ours, analyst at Wikibon. George great to see you. Dave, you know, wrapping up this day what in a phenomenal program. We had a contribution from industry vendors, industry experts, practitioners and customers building and redefining their company's business model. Rolling out technology for Supercloud and multicloud and ultimately changing how they do data. And data was the theme today. So very, very great program. Before we jump into our favorite parts let's give a shout out to the folks who make this possible. Free contents our mission. We'll always stay true to that mission. We want to thank VMware, alkira, ChaosSearch, prosimo for being sponsors of this great program. We will have Supercloud 3 coming up in a month or so, or two months. We'll see. Or sooner, we don't know. But it'll be more about security, but a lot more momentum. Okay, so that's... >> And don't forget too that this program not going to end now. We've got a whole ecosystem speaks track so stay tuned for that. >> John: Yeah, we got another 20 interviews. Feels like it. >> Well, you're going to hear from Saks, Veronika Durgin. You're going to hear from Western Union, Harveer Singh. You're going to hear from Ionis Pharmaceuticals, Nick Taylor. Brian Gracely chimes in on Supecloud. So he's the man behind the cloud cast. >> Yeah, and you know, the practitioners again, pay attention to also to the cloud networking interviews. Lot of change going on there that's going to be disruptive and actually change the landscape as well. Again, as Supercloud progresses to be the next big thing. If you're not on this next wave, you'll drift what, as Pat Gelsinger says. >> Yep. >> To kick off the closing segments, George, Dave, this is a wave that's been identified. Again, people debate the word all you want Supercloud. It is a gateway to multicloud eventually it is the standard for new applications, new ways to do data. There's new computer science being generated and customer requirements being addressed. So it's the confluence of, you know, tectonic plates shifting in the industry, new computer science seeing things like AI and machine learning and data at the center of it and new infrastructure all kind of coming together. So, to me, that's my takeaway so far. That is the big story and it's going to change society and ultimately the business models of these companies. >> Well, we've had 10, you know, you think about it we came out of the financial crisis. We've had 10, 12 years despite the Covid of tech success, right? And just now CIOs are starting to hit the brakes. And so my point is you've had all this innovation building up for a decade and you've got this massive ecosystem that is running on the cloud and the ecosystem is saying, hey, we can have even more value by tapping best of of breed across clouds. And you've got customers saying, hey, we need help. We want to do more and we want to point our business and our intellectual property, our software tooling at our customers and monetize our data. So you have all these forces coming together and it's sort of entering a new era. >> George, I want to go to you for a second because you are big contributor to this event. Your interview with Bob Moglia with Dave was I thought a watershed moment for me to hear that the data apps, how databases are being rethought because we've been seeing a diversity of databases with Amazon Web services, you know, promoting no one database rules of the world. Now it's not one database kind of architecture that's puling these new apps. What's your takeaway from this event? >> So if you keep your eye on this North Star where instead of building apps that are based on code you're building apps that are defined by data coming off of things that are linked to the real world like people, places, things and activities. Then the idea is, and the example we use is, you know, Uber but it could be, you know, amazon.com is defined by stuff coming off data in the Amazon ecosystem or marketplace. And then the question is, and everyone was talking at different angles on this, which was, where's the data live? How much do you hide from the developer? You know, and when can you offer that? You know, and you started with Walmart which was describing apps, traditional apps that are just code. And frankly that's easier to make that cross cloud and you know, essentially location independent. As soon as you have data you need data management technology that a customer does not have the sophistication to build. And then the argument was like, so how much can you hide from the developer who's building data apps? Tristan's version was you take the modern data stack and you start adding these APIs that define business concepts like bookings, billings and revenue, you know, or in the Uber example like drivers and riders, you know, and ETA's and prices. But those things execute still on the data warehouse or data lakehouse. Then Bob Muglia was saying you're not really hiding enough from the developer because you still got to say how to do all that. And his vision is not only do you hide where the data is but you hide how to sort of get at all that code by just saying what you want. You define how a car and how a driver and how a rider works. And then those things automatically figure out underneath the cover. >> So huge challenges, right? There's governance, there's security, they could be big blockers to, you know, the Supercloud but the industry's going to be attacking that problem. >> Well, what's your take? What's your favorite segment? Zhamak Dehghani came on, she's starting in that company, exclusive news. That was big notable moment for theCUBE. She launched her company. She pioneered the data mesh concept. And I think what George is saying and what data mesh points to is something that we've been saying for a long time. That data is now going to flip the script on how apps behave. And the Uber example I think is illustrated 'cause people can relate to Uber. But imagine that for every business whether it's a manufacturing business or retail or oil and gas or FinTech, they can look at their business like a game almost gamify it with data, riders, cars you know, moving data around the value of data. This is something that Adam Selipsky teased out at AWS, Dave. So what's your takeaway from this Supercloud? Where are we in your mind? Well big thing is data products and decentralizing your data architecture, but putting data in the hands of domain experts who can actually monetize the data. And I think that's, to me that's really exciting. Because look, data products financial industry has always been doing building data products. Mortgage backed securities is a data product. But why should the financial industry have all the fun? I mean virtually every organization can tap its ecosystem build data products, take its internal IP and processes and software and point it to the world and actually begin to make money out of it. >> Okay, so let's go around the horn. I'll start, I'll get you guys some time to think. Next question, what did you learn today? I learned that I think it's an infrastructure game and talking to Kit Colbert at VMware, I think it's all about infrastructure refactoring and I think the data's going to be an ingredient that's going to be operating system like. I think you're going to see the infrastructure influencing operations that will enable Superclouds to be real. And developers won't even know what a Supercloud is because they'll be using it. It's the operations focus is going to be very critical. Just like DevOps movements started Cloud native I think you're going to see a data native movement and I think infrastructure is critical as people go to the next level. That's my big takeaway today. And I'll say the data conversation is at the center. I think security, data are going to be always active horizontally scalable concepts, but every company's going to reset their infrastructure, how it looks and if it's not set up for data and or things that there need to be agile on, it's going to be a non-starter. So I think that's the cloud NextGen, distributed computing. >> I mean, what came into focus for me was I think the hyperscaler is going to continue to do their thing, you know, and be very, very successful and they're each coming at it from different approaches. We talk about this all the time in theCUBE. Amazon the best infrastructure, you know, Google's got its you know, data and AI thing and it's playing catch up and Microsoft's got this massive estate. Okay, cool. Check. The next wave of innovation which is coming from data, I've always said follow the data. That's where the where the money's going to be is going to come from other places. People want to be able to, organizations want to be able to share data across clouds across their organization, outside of their ecosystem and make money with that data sharing. They don't want to FTP it anymore. I got it. You take it. They want to work with live data in real time and I think the edge, we didn't talk much about the edge today is going to even take that to a new level real time inferencing at the edge, AI and and being able to do new things with data that we haven't even seen. But playing around with ChatGPT, it's blowing our mind. And I think you're right, it's like when we first saw the browser, holy crap, this is going to change the world. >> Yeah. And the ChatGPT by the way is going to create a wave of machine learning and data refactoring for sure. But also Howie Liu had an interesting comment, he was asked by a VC how much to replicate that and he said it's in the hundreds of millions, not billions. Now if you asked that same question how much does it cost to replicate AWS? The CapEx alone is unstoppable, they're already done. So, you know, the hyperscalers are going to continue to boom. I think they're going to drive the infrastructure. I think Amazon's going to be really strong at silicon and physics and squeeze every ounce atom out of every physical thing and then get latency as your bottleneck and the rest is all going to be... >> That never blew me away, a hundred million to create kind of an open AI, you know, competitor. Look at companies like Lacework. >> John: Some people have that much cash on the balance sheet. >> These are security companies that have raised a billion dollars, right? To compete. You know, so... >> If you're not shifting left what do you do with data, shift up? >> But, you know. >> What did you learn, George? >> I'm listening to you and I think you're helping me crystallize something which is the software infrastructure to enable the data apps is wide open. The way Zhamak described it is like if you want a data product like a sales and operation plan, that is built on other data products, like a sales plan which has a forecast in it, it has a production plan, it has a procurement plan and then a sales and operation plan is actually a composition of all those and they call each other. Now in her current platform, you need to expose to the developer a certain amount of mechanics on how to move all that data, when to move it. Like what happens if something fails. Now Muglia is saying I can hide that completely. So all you have to say is what you want and the underlying machinery takes care of everything. The problem is Muglia stuff is still a few years off. And Tristan is saying, I can give you much of that today but it's got to run in the data warehouse. So this trade offs all different ways. But again, I agree with you that the Cloud platform vendors or the ecosystem participants who can run across Cloud platforms and private infrastructure will be the next platform. And then the cloud platform is sort of where you run the big honking centralized stuff where someone else manages the operations. >> Sounds like middleware to me, Dave >> And key is, I'll just end with this. The key is being able to get to the data, whether it's in a data warehouse or a data lake or a S3 bucket or an object store, Oracle database, whatever. It's got to be inclusive that is critical to execute on the vision that you just talked about 'cause that data's in different systems and you're not going to put it all into some new system. >> So creating middleware in the cloud that sounds what it sounds like to me. >> It's like, you discovered PaaS >> It's a super PaaS. >> But it's platform services 'cause PaaS connotes like a tightly integrated platform. >> Well this is the real thing that's going on. We're going to see how this evolves. George, great to have you on, Dave. Thanks for the summary. I enjoyed this segment a lot today. This ends our stage performance live here in Palo Alto. As you know, we're live stage performance and syndicate out virtually. Our afternoon program's going to kick in now you're going to hear some great interviews. We got ChaosSearch. Defining the network Supercloud from prosimo. Future of Cloud Network, alkira. We got Saks, a retail company here, Veronika Durgin. We got Dave with Western Union. So a lot of customers, a pharmaceutical company Warner Brothers, Discovery, media company. And then you know, what is really needed for Supercloud, good panels. So stay with us for the afternoon program. That's part two of Supercloud 2. This is a wrap up for our stage live performance. I'm John Furrier with Dave Vellante and George Gilbert here wrapping up. Thanks for watching and enjoy the program. (bright music)

>> My name is Franco Nori. Is a great pleasure to be here and I thank you for attending this meeting and I'll be talking about some of the work we are doing within the NTT-PHI group. I would like to thank the organizers for putting together this very interesting event. The topics studied by NTT-PHI are very exciting and I'm glad to be part of this great team. Let me first start with a brief overview of just a few interactions between our team and other groups within NTT-PHI. After this brief overview or these interactions then I'm going to start talking about machine learning and neural networks applied to computationally difficult problems in quantum physics. The first one I would like to raise is the following. Is it possible to have decoherence free interaction between qubits? And the proposed solution was a postdoc and a visitor and myself some years ago was to study decoherence free interaction between giant atoms made of superconducting qubits in the context of waveguide quantum electrodynamics. The theoretical prediction was confirmed by a very nice experiment performed by Will Oliver's group at MIT was probably so a few months ago in nature and it's called waveguide quantum electrodynamics with superconducting artificial giant atoms. And this is the first joint MIT Michigan nature paper during this NTT-PHI grand period. And we're very pleased with this. And I look forward to having additional collaborations like this one also with other NTT-PHI groups, Another collaboration inside NTT-PHI regards the quantum hall effects in a rapidly rotating polarity and condensates. And this work is mainly driven by two people, a Michael Fraser and Yoshihisa Yamamoto. They are the main driving forces of this project and this has been a great fun. We're also interacting inside the NTT-PHI environment with the groups of marandI Caltech, like McMahon Cornell, Oliver MIT, and as I mentioned before, Fraser Yamamoto NTT and others at NTT-PHI are also very welcome to interact with us. NTT-PHI is interested in various topics including how to use neural networks to solve computationally difficult and important problems. Let us now look at one example of using neural networks to study computationally difficult and hard problems. Everything we'll be talking today is mostly working progress to be extended and improve in the future. So the first example I would like to discuss is topological quantum phase transition retrieved through manifold learning, which is a variety of version of machine learning. This work is done in collaboration with Che, Gneiting and Liu all members of the group. preprint is available in the archive. Some groups are studying a quantum enhanced machine learning where machine learning is supposed to be used in actual quantum computers to use exponential speed-up and using quantum error correction we're not working on these kind of things we're doing something different. We're studying how to apply machine learning applied to quantum problems. For example how to identify quantum phases and phase transitions. We shall be talking about right now. How to achieve, how to perform quantum state tomography in a more efficient manner. That's another work of ours which I'll be showing later on. And how to assist the experimental data analysis which is a separate project which we recently published. But I will not discuss today because the experiments can produce massive amounts of data and machine learning can help to understand these huge tsunami of data provided by these experiments. Machine learning can be either supervised or unsupervised. Supervised is requires human labeled data. So we have here the blue dots have a label. The red dots have a different label. And the question is the new data corresponds to either the blue category or the red category. And many of these problems in machine learning they use the example of identifying cats and dogs but this is typical example. However, there are the cases which are also provides with there are no labels. So you're looking at the cluster structure and you need to define a metric, a distance between the different points to be able to correlate them together to create these clusters. And you can manifold learning is ideally suited to look at problems we just did our non-linearities and unsupervised. Once you're using the principle component analysis along this green axis here which are the principal axis here. You can actually identify a simple structure with linear projection when you increase the axis here, you get the red dots in one area, and the blue dots down here. But in general you could get red green, yellow, blue dots in a complicated manner and the correlations are better seen when you do an nonlinear embedding. And in unsupervised learning the colors represent similarities are not labels because there are no prior labels here. So we are interested on using machine learning to identify topological quantum phases. And this requires looking at the actual phases and their boundaries. And you start from a set of Hamiltonians or wave functions. And recall that this is difficult to do because there is no symmetry breaking, there is no local order parameters and in complicated cases you can not compute the topological properties analytically and numerically is very hard. So therefore machine learning is enriching the toolbox for studying topological quantum phase transitions. And before our work, there were quite a few groups looking at supervised machine learning. The shortcomings that you need to have prior knowledge of the system and the data must be labeled for each phase. This is needed in order to train the neural networks . More recently in the past few years, there has been increased push on looking at all supervised and Nonlinear embeddings. One of the shortcomings we have seen is that they all use the Euclidean distance which is a natural way to construct the similarity matrix. But we have proven that it is suboptimal. It is not the optimal way to look at distance. The Chebyshev distances provides better performance. So therefore the difficulty here is how to detect topological quantifies transition is a challenge because there is no local order parameters. Few years ago we thought well, three or so years ago machine learning may provide effective methods for identifying topological Features needed in the past few years. The past two years several groups are moving this direction. And we have shown that one type of machine learning called manifold learning can successfully retrieve topological quantum phase transitions in momentum and real spaces. We have also Shown that if you use the Chebyshev distance between data points are supposed to Euclidean distance, you sharpen the characteristic features of these topological quantum phases in momentum space and the afterwards we do so-called diffusion map, Isometric map can be applied to implement the dimensionality reduction and to learn about these phases and phase transition in an unsupervised manner. So this is a summary of this work on how to characterize and study topological phases. And the example we used is to look at the canonical famous models like the SSH model, the QWZ model, the quenched SSH model. We look at this momentous space and the real space, and we found that the metal works very well in all of these models. And moreover provides a implications and demonstrations for learning also in real space where the topological invariants could be either or known or hard to compute. So it provides insight on both momentum space and real space and its the capability of manifold learning is very good especially when you have the suitable metric in exploring topological quantum phase transition. So this is one area we would like to keep working on topological faces and how to detect them. Of course there are other problems where neural networks can be useful to solve computationally hard and important problems in quantum physics. And one of them is quantum state tomography which is important to evaluate the quality of state production experiments. The problem is quantum state tomography scales really bad. It is impossible to perform it for six and a half 20 qubits. If you have 2000 or more forget it, it's not going to work. So now we're seeing a very important process which is one here tomography which cannot be done because there is a computationally hard bottleneck. So machine learning is designed to efficiently handle big data. So the question we're asking a few years ago is chemistry learning help us to solve this bottleneck which is quantum state tomography. And this is a project called Eigenstate extraction with neural network tomography with a student Melkani , research scientists of the group Clemens Gneiting and I'll be brief in summarizing this now. The specific machine learning paradigm is the standard artificial neural networks. They have been recently shown in the past couple of years to be successful for tomography of pure States. Our approach will be to carry this over to mixed States. And this is done by successively reconstructing the eigenStates or the mixed states. So it is an iterative procedure where you can slowly slowly get into the desired target state. If you wish to see more details, this has been recently published in phys rev A and has been selected as a editor suggestion. I mean like some of the referees liked it. So tomography is very hard to do but it's important and machine learning can help us to do that using neural networks and these to achieve mixed state tomography using an iterative eigenstate reconstruction. So why it is so challenging? Because you're trying to reconstruct the quantum States from measurements. You have a single qubit, you have a few Pauli matrices there are very few measurements to make when you have N qubits then the N appears in the exponent. So the number of measurements grows exponentially and this exponential scaling makes the computation to be very difficult. It's prohibitively expensive for large system sizes. So this is the bottleneck is these exponential dependence on the number of qubits. So by the time you get to 20 or 24 it is impossible. It gets even worst. Experimental data is noisy and therefore you need to consider maximum-likelihood estimation in order to reconstruct the quantum state that kind of fits the measurements best. And again these are expensive. There was a seminal work sometime ago on ion-traps. The post-processing for eight qubits took them an entire week. There were different ideas proposed regarding compressed sensing to reduce measurements, linear regression, et cetera. But they all have problems and you quickly hit a wall. There's no way to avoid it. Indeed the initial estimate is that to do tomography for 14 qubits state, you will take centuries and you cannot support a graduate student for a century because you need to pay your retirement benefits and it is simply complicated. So therefore a team here sometime ago we're looking at the question of how to do a full reconstruction of 14-qubit States with in four hours. Actually it was three point three hours Though sometime ago and many experimental groups were telling us that was very popular paper to read and study because they wanted to do fast quantum state tomography. They could not support the student for one or two centuries. They wanted to get the results quickly. And then because we need to get these density matrices and then they need to do these measurements here. But we have N qubits the number of expectation values go like four to the N to the Pauli matrices becomes much bigger. A maximum likelihood makes it even more time consuming. And this is the paper by the group in Inns brook, where they go this one week post-processing and they will speed-up done by different groups and hours. Also how to do 14 qubit tomography in four hours, using linear regression. But the next question is can machine learning help with quantum state tomography? Can allow us to give us the tools to do the next step to improve it even further. And then the standard one is this one here. Therefore for neural networks there are some inputs here, X1, X2 X3. There are some weighting factors when you get an output function PHI we just call Nonlinear activation function that could be heavy side Sigmon piecewise, linear logistic hyperbolic. And this creates a decision boundary and input space where you get let's say the red one, the red dots on the left and the blue dots on the right. Some separation between them. And you could have either two layers or three layers or any number layers can do either shallow or deep. This cannot allow you to approximate any continuous function. You can train data via some cost function minimization. And then there are different varieties of neural nets. We're looking at some sequel restricted Boltzmann machine. Restricted means that the input layer speeds are not talking to each other. The output layers means are not talking to each other. And we got reasonably good results with the input layer, output layer, no hidden layer and the probability of finding a spin configuration called the Boltzmann factor. So we try to leverage Pure-state tomography for mixed-state tomography. By doing an iterative process where you start here. So there are the mixed States in the blue area the pure States boundary here. And then the initial state is here with the iterative process you get closer and closer to the actual mixed state. And then eventually once you get here, you do the final jump inside. So you're looking at a dominant eigenstate which is closest pure state and then computer some measurements and then do an iterative algorithm that to make you approach this desire state. And after you do that then you can essentially compare results with some data. We got some data for four to eight trapped-ion qubits approximate W States were produced and they were looking at let's say the dominant eigenstate is reliably recorded for any equal four, five six, seven, eight for the ion-state, for the eigenvalues we're still working because we're getting some results which are not as accurate as we would like to. So this is still work in progress, but for the States is working really well. So there is some cost scaling which is beneficial, goes like NR as opposed to N squared. And then the most relevant information on the quality of the state production is retrieved directly. This works for flexible rank. And so it is possible to extract the ion-state within network tomography. It is cost-effective and scalable and delivers the most relevant information about state generation. And it's an interesting and viable use case for machine learning in quantum physics. We're also now more recently working on how to do quantum state tomography using Conditional Generative Adversarial Networks. Usually the masters student are analyzed in PhD and then two former postdocs. So this CGANs refers to this Conditional Generative Adversarial Networks. In this framework you have two neural networks which are essentially having a dual, they're competing with each other. And one of them is called generator another one is called discriminator. And there they're learning multi-modal models from the data. And then we improved these by adding a cost of neural network layers that enable the conversion of outputs from any standard neural network into physical density matrix. So therefore to reconstruct the density matrix, the generator layer and the discriminator networks So the two networks, they must train each other on data using standard gradient-based methods. So we demonstrate that our quantum state tomography and the adversarial network can reconstruct the optical quantum state with very high fidelity which is orders of magnitude faster and from less data than a standard maximum likelihood metals. So we're excited about this. We also show that this quantum state tomography with these adversarial networks can reconstruct a quantum state in a single evolution of the generator network. If it has been pre-trained on similar quantum States. so requires some additional training. And all of these is still work in progress where some preliminary results written up but we're continuing. And I would like to thank all of you for attending this talk. And thanks again for the invitation.

>>from around the globe. It's the Cube with digital coverage of next level network experience event brought to >>you by info blocks. Okay, welcome back, everyone. This is the Cube's coverage of the next level networking experience. Virtual event within four blocks. I'm John Furrow, your host of the Cube. We're here in our Palo Alto, Calif. Studios as part of our remote access during Covic, getting the interviews and the stories and sharing that with you. We got a great guest here, then savages the network operations manager at Morgan School District in Utah. A customer of info blocks to share a story. Then thanks for coming on. >>Thanks for having >>me. First of all, the Red Sox had a plus interview. I would say right now is gonna go great. Go Sox. Which baseball was in season. Great to have you on. Um, >>we'll get there. We'll >>get there. Um, my Yankee fans say when I say that. But anyway, Miss baseball, um, you know. But that brings up covert 19 baseball season sports. Life has been impacted. Your district. Like many school districts around the world, we're told to shut down, send workers home. That meant sending kids home, too. So we got the educators, get the administration, and you've got the kids all going home. >>Yeah. >>What did you do to keep things going? Because then stop. They had to do the remote learning and new things were emerging. New patterns, new traffic, new kinds of experiences. What did you learn? What's going on? >>Well, first we tried to lock the doors and pretend we weren't there, but they found us. Um, really? I mean, real quickly in our school district, we're not a 1 to 1 operation, so the, uh that caused a big change for us. Um, we had to quickly adapt. And we chose to use chromebooks because that's what we have for the students to use in their classes. So getting that, uh, squared away and send out into the family's was was a big challenge. But then on top of that being the school district, we then had to decide. Okay, how do we protect and filter provide the filtering that the students are gonna need even though they're at home? So there's some relative safety there when they're online and and accessing your email and things like that. So those were. Our two are probably our two. Biggest hurdles was, you know, ramping up the devices and then and then providing, making sure, you know, the network access from a filtering and consistency standpoint was going to work. >>You know, I got to ask you because I see this kind of disruption you don't You don't read about this in the i t. Manual around disaster recovery and, you know, disruption to operations. But essentially, the whole thing changes, but you still got to connect to the network, DNS. You gotta get the access to the content. You got content, you get systems. You got security all to be managed while in flight of dealing with connection points that remote. So you've got the disruption and the craziness of that, and then you've got this big I o t experiment basically edge of the network, you know, in all over the place. You know, on one hand, you kind of geek out and say, Wow, this is really kind of a challenge is an opportunity to solve the problem at the same time, you know, What do you do? So take us through that because that's a is a challenge of locking down the security in a borderless environment. People are everywhere. The students business has to get done. You got to resolve to. The resource is >>so thankfully, we had migrated If it blocks several years ago. Um and just this last, I would say in October, I finally got us on. Ah, cloud the blocks. One threat defense Cloud portion of it too. So from a security standpoint, we already had a really good, um foundation in place from both the DNs aspect and the DNS security aspect. Um so that was to be honest, most users. It was seamless transition. In many regards, both users didn't even realize they were being, You know, pushed through the info blocks is cloud DNs server, you know, which was providing security and filtering. So that was a big plus for us because it it was less man hours. We had to spend troubleshooting people's DNS resolutions. Why sites Wouldn't you know? Maybe they weren't being filtered correctly. All that was was to be honest, perfect. Where other platforms we had previously were just a nightmare to manage, >>like, for example, of the old way versus the new way here and marital, is it? What files configuration will take us through? What? You >>know, it was like a separate. It was a separate product content filter that works in conjunction with the firewall. Um, and I'm not going to name the company's name. I don't want, you know, even though many company but it seemed with that product we were spending, on average about 3 to 4 hours a day fixing false positives just from a filtering aspect because it would interfere with the DNS. And it does. It didn't really do it. I mean, how it filters is not based on DNS. Totally right. So by migrating temple blocks are DNS and the filtering the security is all handling at the DNs level. And it was just much more, um, to be I mean, frankly, honestly, is much more invisible to the end user. So >>more efficient. You decouple filtering from DNs resolution. Got it. All right, this is the big topic. I've been talking with info blocks people on this program in this event is on how this new d d I layer DNs d XP and I p address management kind of altogether super important. It's critical infrastructure Yeah. No spoilers, Enterprise. You're borderless institution. Same thing you go to school as a customer. How does the d I lay out this foundational security play for delivering this next level experience? What's your take on that? >>Well, for our like, for a school platform, we we use it in a number of ways. Besides, I mean, the filtering is huge, but just for the ability, like, for example, one of the components is is response policy zones or DNS firewalls what they call it, and that allows you one to manage, um, traditional, like DNS names, right? P addresses you can. You can manage those by creating essentially a zone that is like a white list of blacklist rewrite. So you've got a lot of control, and again it's filtering at the DNs level, so it's looking based on DNS responses inquiry. The other aspect of that is, is the feeds that you receive from info blocks. So by subscribing to those, we, um we have access to a lot of information that info Blocks and their partners have created identifying, you know, bad actors, malware attack vectors based on again DNs, uh, traffic, if you will, and so that takes a load office. Not having to worry. I'm trying to do all that on our own. I mean, we've seen a lot of attacks minimized because of the feeds themselves. So that again frees us up. We're a very small school district. In some regards, there's a I am the only network person in the district, and there's like, a total of four of us that manage, you know, kind of the support aspect. And so, being able to not have to spend time researching or tracking down, you know, breaches and attacks as much because of the DNS. Security frees me up to do other things, you know, like in the more standard networking realm, from a design and implementation. >>Great. Thanks for sharing that. I want to ask about security as a very competitive space security here and everyone promising it different things at different security things. You know, by I gotta ask you, why did you guys decide to use info blocks and what's the reason behind it? >>Well, to be frankly honest, I'm actually in info blocks trainer and I've been training for 15 years, so I kind of had an agenda when I first took this job to help out the school district. In my experience, I've been doing working in networking for over 20 years. And in my experience, I ever boxes one of the most easy and in best managed DNS solutions that I've come across. So, um, you know, I might be a little biased, but I'm okay with that. And so I I pushed us to be honest, to get there and then from the security aspect has all that has evolved. It just makes to me it makes sense. Why not wrap the more things you can maybe wrapped together. And so you know, when you're talking about attacks, over 90% of attacks use DNS. So if I have a solution that is already providing my DNS and then wraps the security into it, it just makes the most sense for me. >>Yeah. I mean, go back. The info box is DNA. You got cricket. Liu Stuart Bailey, the founder, was this is zero. This didn't just wake up one day and decided to start up these air practitioners early days of the Internet. They know DNS cold and DNS is we've been evolved. I mean, and when it needs that when you get into the DNS. Hacks and then you realize Okay, let's build an abstraction layer. You've seen Internet navigation discovery, all the stuff that's been proven. It is a critical infrastructure. >>Well, and to be honest, it's It's one of those services that you can't can't filter the firewall right. You have to have it. You have to. It's that foundation layer. And so it makes sense that Attackers air leveraging it because the fire will has to let it through in and out. And so it's a natural, almost a natural path for them to break in. So having something that speaks native DNS as part of your security platform makes more sense because it it can understand and see those attacks, the more sophisticated they become as well. >>So I gotta ask you, since you're very familiar info blocks and you're actually deploying its great solution. But I got this new DD I Layer, which is an abstraction, is always a great evolution. Take away complexity and more functionality. Cloud certainly cloud natives everywhere. That's but if it's for what is the update, if if I'm watching this month, you know I've been running DNS and I know it's out there. It's been running everything. And I got a update, my foundation of my business. I got to make my DNS rock solid. What's the new update? What's info blocks doing now? I know they got DNS chops seeing that on it. What's new about info blocks? What do you say? >>Well, it's, you know, they have a couple things that they've been trying to modify over the last several years. In my opinion, making more DNS like a you know, like software as a service, you know, service on demand, type of approach. That's a yes. So you have the cloud components to where you can take a lot of the heavy lifting, maybe off of your network team's shoulders. Because it is, it is. Um, I think people will be surprised how many customers out there. I have, ah, teams that are managing the DNS and even the D HCP aspect that that's not really what their experiences and then they don't They don't have, ah, true, maybe background Indians, and so having something that can help make that easier. It's almost, you know, hey, maybe used this term it almost sounds like it's too simple, but it's almost like a plug and play approached for some. For some environments, you know you're able to pop that in, and a lot of probably the problems they've been dealing with and not realizing what the root cause was will be fixed. So that's always a huge component with with info blocks. But their security is really what's come about in the last several years, Um, and and back as a school district, you know, our besides securing traffic, which every customer has to do, um, we have our you know, we're We have a lot of laws and regulations around filtering with with students and teachers. So anyone that's using a campus own device And so for us this I don't think people realized that the maturity that the filtering aspect of the blocks one defence now it's it's really evolved over the last couple of years. It's become a really, really good product and, like I said earlier, just work seamlessly with the data security. So it is going to be using >>an SD Wan unpacked everything. You go regular root level DNs is it? So I gotta ask you. How is the info blocks helping you keep network services running in system secure? >>Well, I think I think we're more on just the DNs d It does R d eight DNS and DCP. So from that standpoint, you know, in the five years almost we've been running that aspect. We have had very little if if maybe one or two incidents of problems with, you know from a DNS TCP so so are our users are able to connect, you know, when they turn on their computer To them, the Internet's up. You know, there's no there's no bumps in the road stopping them from from being able to connect. So that's a huge thing. You know, you don't have to deal with those Those constant issues again is a small team that just takes time away from the big projects. You're trying to, um, and then to the being able to now combine things. Security filtering solution. Uh, that alone has probably saved us. Oh, we'll probably you know, upwards of 500 man hours in the last eight months. So where normally we would be spending those hours again, troubleshooting issues that false positives, things like that. And there's a small team that just sucks the life out of you when you have to. You always spend time on that. >>I mean, you always chasing your tails. Almost. You want to be productive. Automation plays >>a >>key role in that, >>right? Yeah. >>So I got to ask you, you know, just a general question. I'm curious. You know, one of the things I see is sprawling of devices. WiFi was a great example that put an access point up a rogue access point, you know, as you get more connections. De HCP was amazing about this is awesome. But also, you had also de HCP problem. You got the the key Management is not just around slinging more d HDP around. So you got the trend? Is more connections on the eyepiece? Not how does info blocks make that easier? Because for people who may not know, the DNS ends announcing TCP and IP address management. They're all kind of tied together. Right? So this >>is the >>magic of DD I in my head. I want to get your thoughts on how you see that. Evolving. >>Yeah, I think that's another kind of back twice. It's kind of almost like a plug and play for a lot of customer environments. They're getting, you know, you're getting the DSP, DNs and eye Pam all wrapped in once you have this product that speaks, well, those languages, if you will and that And, um along with some of the reporting services and things of that nature. Um, when I look for, like, a Mac address in my influx database, I'm not just going to get ah, Mac address and what the i p addresses. I'm not just going to get the DNs like the host name. Maybe you know, the beauty and fully qualified domain name. Either I have the ability to bring in all this information that one. The client is communicating with the DCP DNS server on top of things like metadata that you can configure in the database to help really color in the picture of your network. So when you're looking at what device is using this I p when we talk about rogue devices or things like that, uh, I can get so much more information out of info blocks that almost almost to the point where you're almost being able to nail down the location of where the devices that even if it's a wireless client because it works in conjunction with some of our wireless appointments, too. So within, you know, a matter of minutes we have almost all the information we would need to take whatever action is appropriate for something like that, that getting used to take us hours and hours to troubleshoot. >>Appreciate a lot of the other interviews I've done with the info blocks, folks. One of the things that came out of them is the trailing. You can see the trail they're getting. They got to get in somewhere. DNS is the footprints of there you got? That's the traffic, and that's been helping on a potential attacks in D DOS is, for example, no one knows what that is, but DNS is what he said. A lot of the surface areas, DNS. With the hackers are makes it easier to find things. >>Well, you know, by integrating with the cloud I've I've got, you know, that the cloud based with the blocks one, it added a advanced DNS security, which helps protect skins Adidas as well as any cast to help provide more availability because I'm pushing on my DNs traffic through those cloud servers. It's like I've I'm almost equivalent of a very large organization that would normally spend millions of millions of dollars trying to do this on their own. So I'm getting the benefits and kind of the equivalent from that cloud hybrid approach that normally we would never have have. The resource is, >>Well, then I really appreciate you taking the time out of your busy day to remote into the Cube studios. Talk about next level networking experience, so I want to just ask you, just put your experience hat on. You've been You've seen some waves. You've seen the technology evolve when you hear next level networking and when you hear next level networking experience almost two separate meetings. But next level networking means next level. Next level networking experience means is some experience behind it. One of those two phrases mean to you next level networking and next level networking experience. >>Well, to me, I always look at it as the evolution of being able to have a user experience that's consistent no matter where you're located, with your home in your office and special with in today's environment. We have to be able to provide that consistent experience. But what I think what a lot of people may not think about or my overlook if you're just, you know, more of an end user is along with that experience, it has to be a consistent excess security approach. So if I'm an end user, um, I should be able to have the access the, um and the security, which, you know, you know, filtering all that fun stuff to not just allow me the connectivity, but to bring me, you know, that to keep the secure wherever I met. And ah, um, I think schools, you know, obviously with code and in the one the one that everyone was forced to do. But I think businesses And generally I think that's, you know, years ago, Cisco when I worked with Cisco, we talked about, you know, the remote user of the mobile user and how Cisco is kind of leading, uh, the way on that. And I think, you know, with the nature of things like this pandemic, I think being able to have your your users again have that consistent experience, no matter where they're at is going to be key. And so that's how I see when I think of the network evolution, I think that's how it it has to go. >>Well, we appreciate your your time sharing your insights Has a lot of a lot of people are learning that you've got to pour the concrete to build the building. DNS becoming kind of critical infrastructure. But final question for you. I got you here, you know? How you doing? Actually, schools looks like they're gonna have some either fully virtual for the next semester or some sort of time or set schedule. There's all kinds of different approaches. This is the end of the day. It's still is this big i o t experiment from a traffic standpoint. So new expectations create new solutions. What do you see on the horizon? What challenges do you see as you ride this way? Because you've got a hold down the fort, their school district for 3000 students. And you got the administration and the faculty. So you know What are you expecting? And what do you hope to see Evolve Or what do you want to stay away from? What's your opinion? >>I think? I think my my biggest concern is, you know, making sure our like, our students and staff don't, uh, you know, run into trouble on by say that more from, you know, you know, by being, you know, being exposed to attacks, you know, their data with Delta becomes, you know, comes back to our data as a district. But, you know, the student data, I think I think, you know, with anything kids are very vulnerable. Ah, very role, vulnerable targets for many reasons. You know, they're quick to use technology that quick to use, like social media, things like that. But they're they're probably the first ones to do security Does not, you know, across their mind. So I think my big my big concern is as we're moving this, you know, hybrid, hybrid approach where kids can be in school where they're going to be at home. Maybe they'll change from the days of the week. It'll fluctuate, uh, keeping them secure, you know, protecting them from themselves. Maybe in a way, if I have to be the guy is kind of the grumpy old dad it looked at. I'm okay with wearing that hat. I think that's my biggest. Our concern is providing that type of, uh, stability and security. So parents at the end of that could be, you know, I have more peace of mind that their kids you know, our online even more. It's great >>that you can bring that experience because, you know, new new environments, like whether it zooming or using, try and get the different software tools that are out there that were built for on premise premises. You have now potentially a click here. Click there. They could be a target. So, you know, being safe and getting the job done to make sure they have up time. So the remote access it again. If you've got a new edge now, right? So the edge of the network is the home. Exactly. Yeah. Your service area just got bigger. >>Yeah. Yeah, we're in. You know, I'm everybody's guest, whether they like it or not. >>I appreciate that. Appreciate your time and good luck. And let's stay in touch. Thanks for your time. >>Hey, thanks for having me. You guys have a good rest of your weekend? Day two. State State. >>Thank you very much. It's the Cube's coverage with info blocks for a special next level networking experience. Pop up event. I'm John for the Cube. Your host. Thanks for watching. Yeah, yeah, yeah.

>>from around the globe. It's the Cube with digital coverage of next level network experience event brought to you by Info blocks Hi ups to Minuteman and welcome to the Cube's coverage of the info blocks virtual event. Digging into the next level networking experience. I'm here with John Furrier, who is the host of the event. John. We've been talking about next level networking for for a few years now. Everything's multi cloud cloud native SAS adoption, really transforming the way that we have to think about networking. Tell us a little bit about this event. >>So as you know, yeah, again go back years from when member VM Ware bought in a sexual like Okay, you know that's going to change the game software to find networking. And we love that. We were all riffing on program ability. You saw the Dev Ops trajectory hitting networking. We would say that's where the action is on this event really kind of speaks to Info Blocks as a company which is really well known for DNS. I mean, they had cricket. Liu Stuart Bailey, that really kind of the pioneers in DNS and security have constantly been adding innovation to it, but DNS is one of those things where it's kind of like not thought about, but it runs everywhere, runs the web. It is critical infrastructure and, you know D HCP. We all know what that is. We have a home router, and then he got I p address management. These have been traditionally different things for enterprises, and everyone has it. They got to deal with it. And it's really, ultimately the location and how things resolved and connect. So you know, it really becomes a foundational opportunity to figure out where the access is not only a remote access, but security. So we had a great bunch of guests looking at looking at the info blocks. Next level networking, because they bought, had an acquisition, a Cube alumni snap route recently, and this caught our attention because they were doing Cloud Native. And one of the guests we had was Glenn Sullivan. He was the founder of Snap Route. He was the the guy who did all the Siri work for Apple. So this guy knows large scale of those cloud native We had kuna Sunni, who's the runs? Corporate development in all of the products for info blocks. He kind of went into the strategy of how they're taking the I won't say boring DNs, but the critical infrastructure of DNS and how they're extending the functionality with an abstraction layer around D D I, which is DNS DCP and management. And then we had some great guests on there. We had a Craig Sanderson from info blocks. He's on there. You'll hear from him. He talked about the security and then finally a customer who's running a big school district who, with Covert 19 exposes all these challenges around what has been called the borderless enterprise. So really, next level is that, you know, how do you deal with all this stuff? And that's been a big issue. So we're gonna unpack all that in this virtual event. We have four great interviews, and so it's going to be a great program. >>Yeah, John, as you said it to some of those foundational pieces of how network is done, a lot of times runs, you know, under the radar, something you don't need to think about. But all of these changes, as we said, you know my data. My network is now highly distributed, especially I would expect that the impact of the global pandemic and work from home are really causing even more of these challenges and to think about distributed infrastructure even more. So what are some of the themes we should be looking for here? How much of them kind of tie into what we've been talking about the last couple of years in some of these cloud native worlds? >>That's great questions to I'll get into some of the themes of the program, but you brought up the covert 19 again. We've been talking about this in our reporting. You've been doing a ton of interviews following all your your stuff as well as well as all of our team. Covert 19 really exposes the aspect of critical infrastructure, and to me it's like it's the It's the great I o T experiment happening in real time. It's forcing companies saying, Hey, the work. The future of work is about workplace. The location is now home workforce. Are the people emotional? They want ease of use. They want a different experience. They're all not in the office workloads and work flows. All of them have the common word working it so I think over 19 exposes this what I call I o t experiment because everyone is now borderless. It changes the game and really puts the pressure on security network access. And ultimately, you know, the bad guys are out there so you could have someone a teacher at home or a worker at home, and they get some malware attack and they're not sophisticated, zoom or whatever they're using for tools. All that's changed and they're vulnerable. So this brings up a huge networking challenge from whether even VP ends or even relevant or not to everything. So, to me, that is a huge point. You're gonna hear that throughout the commentary that that's kind of teased out. But the real things about innovation around the cloud you're gonna hear info blocks and they're experts talk about what they're doing and how they see cloud scale and cloud native integrating into an older paradigm like DNS. And to me, that's the That's the evolution of this DD I concept. That's an abstraction layer that creates innovation opportunities but also takes away a lot of the complexities around managing all the DNS things out there and again, that's the access of the network. It's a it's a place of truth is really kind of low level, but it's really foundational. So to me, that's the main theme. And customers want ease of use into it, whether they're at home or not, and replacing the old ways to putting a box out there. That's the way it was, DNs DNs. People would manage it all. Now they want to have it provisioned, managed a manage service cloud Native Cloud operations because it's only gonna get has to get that way. >>Yeah, it's interesting, John. You know, we watched the whole wave of software defined impact networking. I think of a company like Info blocks. They've been around for decades. They're dominant in the space is that they play in. Traditionally, it would have been an appliance that you thought of for their environment you talked about. They now have the snapper out acquisition as part of what they're doing. So it just what should we be looking for? What are they really the main point? That Info Box wants to bring people together for this next level networking experience? >>Well, Glenn Sullivan was one of my favorite discussions, and he's been on. He's a cube alumni and he's so smart. He came again from Apple. He knows that he knows what large scale looks like. Snap route was really early and was one of those technologies that just, you know, it has the core DNs built in kubernetes built in. They were doing some pretty aggressive, I would call it for lack of a better word kubernetes on bare metal. They were doing stuff, but really super cool kubernetes you combine that with DNS and info blocks actually has the core DNs that's actually in every kubernetes of in the CN CF. So everything that comes out of the CN CF from a core DNS standpoint is info blocks. So yeah, they're definitely relevant in the whole CNC of Cloud Native foundation, effort around cloud native. And as that scales just micro services, you're gonna have to have this new abstraction layer and also be compatible with automation. So that's, um, we didn't go into the weeds on that, but that was essentially the head room for all the different conversations roles of cloud native and open source technologies enabling borderless enterprises because you got to have the operation side and you got to have the program ability. So you start to get into the true dev ops that we used to riff on all the time. You know, move fast, break stuff to don't break anything. Right? So ops, ops and Dev have to come together. This is where the winners and losers of networking will be determined. You gotta provide the enablement for developers, but you gotta provide the stability of an operational checklist. >>Yeah, John, I guess the last question I want to ask you before we get to the guests, You know, that operational change, You know, we saw for so many years it was, Oh, all the networking people, they're going to have to learn to code up weight. Dev ops is actually gonna spreading the information around. And maybe I won't need a particular networking team. But we understand when things go wrong, you've gotta have somebody with the expertise that could be able to dig in. What are you know, who should be listening to this? What are some of those organizational implications for what you're talking about with info blocks? >>That's a great point. I mean, the biggest challenge that I see in all this entire digital transformation as it starts to get down into the cloud native world is, most people are asking the wrong questions. They don't even know what they're talking about When it comes down to trying to compare an apple to an orange, they're really kind of disconnected on language. You got server people in networking. We know that they have different languages, and working together is key. When you think about something like DNS, that's a technical. That's an operator that's an I t person, that someone who's running critical infrastructure. But when you start to think about the security aspect of it, it's a CSO conversation. So what I'm seeing come out of this that's critical, is when you start to get into this cloud native world. You have more stakeholders in the value proposition of all this and with covert 19. As I pointed out, you know you got hacks and you got security. So when you talk with security, that's up and down the organization. That's the CSO down to the teams themselves. We have about automation horizontally scaling with Dev ops. That's multiple teams, so you have an integration kind of stakeholders. You know DNS servers, all networking. All these people have to kind of come together. So the people who should watch this are the people who are concerned about scaling the modern enterprise, which is borderless, which is code word for multiple access points and multiple connection points. R i o t um, how do you make that work? And that's the real challenge. So it's kind of like an I t a person who wants to figure out where the puck will be so they could be there when it's there and skate to where the puck is, as we say, and and the CSO of the senior people have to understand that DNS cannot be overlooked because whether it's a managed service. So So Cloudflare had a huge out into the DNS. Setting DNS takes down everything. So it's ah, it's the most fertile ground and the most targeted ground for attacks, and that is well understood. So getting the right questions in place foundational we had to set up the modern enterprise, and then that's going to be a construct for the cloud native architecture and ultimately the developer environment. So yeah, it's a topic that's kind of nerdy with DNS, But it has implications across digital transformation. >>Jonah expecting lots of conversations around security and automation how they tie into all of the modern and modernization themes. Absolutely some pieces that shouldn't be left behind. All right, John Ferrier, Thanks so much for helping us kick off. Really interested. Make sure to stick with us off to listen to all the guest interviews here that John has done the info blocks. Next level networking experience. Instrument, man. And thank you for watching the Cube. Yeah, yeah, yeah, yeah, yeah, yeah.

>> Announcer: SiliconANGLE Media presents theCUBE! covering Alibaba Cloud's annual conference, brought to you by Intel. Now here's John Furrier.... >> Hi I'm John Furrier, with SiliconANGLE, Wikibon and theCUBE. I'm the co-founder based in Silicon Valley, California, Palo Alto, California, and I am here in Hangzhou, China for the Alibaba Cloud conference in Cloud City, it's the biggest cloud computing conference here in China. I'm excited to be here with Dr. Min Wanli, who's the Chief Data Scientist and General Manager of Big Data division at Alibaba Cloud. Dr. Wanli, thank you for spending time. >> Thank you for having me. >> We have seen a lot of data in the conversation here at the show, data technology's a big part of this new revolution, it's an industrial revolution that we've never seen before, a whole 'nother generation of technology. What does data technology mean to Alibaba? >> Okay, it means everything. So first off, our internal technical speaking, it's technology handling massive real-time data and streaming data, and that's of different variety. For instance the app for the mobile app, for system knock, the customer behavior, they click, and they click browsing of the digital image of each merchant and asking for the price and compare against another similar product. All these behaviors are translated as data, and this data will be further merged with the archived data and try to update the profile of this customer's interests, and then try to detect whether there's a good match of they current merchant with the customer intent. If the match is good, then will flash this to the top priority, the top spot. So that try to increase the conversion rate. So if the conversion rate is high, then our sales is high. So DT, data technology means everything to Alibaba. >> It's interesting, I find my observation here, it's so fascinating because in the old days, applications produced data, that was stored on drives. They'd go to data warehouses, and they'd analyze them. You guys, in Alibaba Cloud are doing something fundamentally different, that's exciting in the sense that you have data, people call it data exhaust or data in general, but you're reusing the data in the development in real-time. So it's not just data exhaust, or data from an application. You're using the data to make a better user experience and make the systems smarter and more intelligent. Did I get that right? >> Exactly, exactly. This is a positive feedback loop, in a way, so in the old-fashioned way, you archived the data for offline analysis and for post-event analysis, and trying to identify whether there's any room for improvement. But that's fine. But now people cannot wait, and we cannot wait. Offline is not enough. So we have to do this in real time, online, in a feedback version, in search of a way that we could capture exactly at the right moment, understand the intent of the customer, and then try to deliver the right content to the customer on the fly. >> Jackie Ma, or Jack Ma, your boss, and also Dr. Wong who I spoke with yesterday, talk about two things. Jack Ma talks about a new revolution, a new kind of industrial revolution, a smarter world, a better society. Dr. Wong talks about data flowing like a river, and you hear Hangzhou as an example, but it highlights something that's happening across the world. We're moving from a batch, slow world with data to one that's in motion and always real time. They're not necessarily mutually exclusive, but they're different. A data lake or a data river, whatever word you want, I don't really like the word data lake personally, I think it means, it's batch to me. But batch has been around for a while. Real time mixed streaming. This is something that's happening, and it's impacting the architecture and the value proposition of applications, and it's highlighted in Internet of Things, it's highlighted in examples that we're seeing that's exciting like the ET Brains. Can you share your view in your project around ET Brains, because that is not just one one vertical. It's healthcare, it's industrial, it's transportation, it's consumer, it's everything. >> Yeah, good question, so first of all I concur with you that data lake already exists, will continue to exist, because it's got its own value because our ET Brain for example, actually emerged from data lake, because it has to learn all the benchmark, the baseline model, the basic knowledge from the existing archive data, which is a data lake. However, that's not enough. Once you have the knowledge, you have the capability but you need to put that in action. So we are talking about data in motion, data in action. How do we do that? So once you have the training example, all the training data from data lake, and you train the brain, the brain is mature enough and in the next step you want to push the brain coupled with real-time streaming data, and then to generate real-time action in real-time manner in preemptive way, rather than posting in a reactive way. So for example, in transportation and travel, T and T, travel and transportation, and traffic management. So currently, all the authorities, they have access to real-time information, and then they do a post-event analysis if there's a traffic jam, and then they want to do some mitigation. However, the best scenario is, if you can prevent the traffic jam from happening in the first place, right, how can you foresee there will be, there would be, there could be traffic jam happen in 10 minutes from now, and then you take a preemptive strike, and then try to prevent that from happening. That's the goal ET Brain, in traffic management want to achieve. Like for example, you see the ambulance case, and once the ET Brain receives the message say the ambulance is going to go to Point A, pick up a patient, and carry that patient, rush them to Hospital B, and then it immediately calculates the right routing, the driving direction, and the calculate the ETA to every intermediate intersection and then try to coordinate with the traffic lights, traffic signal. All this systematic integration will create on demand a green wave for ambulance, but in the past ambulance is just by the siren, right. >> Yeah, this is fascinating, and also I'd like to get your thoughts, because you bring us something that's important, and that is, and I'd like to connect the dots for the audience, and that is, real time matters. If you're crossing the street, you can't be near real time, because you could get hit by a car. But also latency's important, also the quality of the data is good. I was talking to an executive who's laying out an architecture for a smart city, and he said, "I want the data in real time," and the IT department said, "Here it is, "it's in real time", and he says, "No, that's last year's data." And so the data has to be real time and the latency has to be low. >> Exactly. I completely agree. The latency has to be low. Unfortunately, in the current IT infrastructure, very often the latency exist. You cannot eliminate that, right? And then you have to live with that, so the ET Brain acknowledge the fact, in fact we have our own algorithm designed in a way that it can make a shortened prediction. So based on five minutes ago data, the data collected five minutes ago, and then it can project the next five minutes, next 10 minutes, what would be the data, and then use that to mitigate, or to conquer, to offset the latency. So we find that to be a good strategy, because it's relatively easy to implement, and it is fast and efficient. >> Dr. Wanli, fascinating conversation. I'd like to get your thoughts on connecting that big data conversation or data conversation to this event. This is a cloud computing event. We at theCUBE and SiliconANGLE and our Wikibon research team we go to all the events. But sometimes the big data events are about big data, Hadoop, whatever, and then you have cloud, talking about DevOPs, and virtual machines. This conference is not just a siloed topic. You have cloud computing, which is the compute, it's the energy, it's the unlimited compute potential, but it's also got a lot of data. You guys are blending it in. >> Exactly. >> Is that by design, and why is that important? >> It's by design. Actually, you cannot separate cloud from data, big data. Or you cannot talk big data without referring to cloud, because once the data is big, you need a huge computation power. Where does that come from? Cloud computing. So that means that data intelligence, all the value has to require a good technological tool to unleash the value. What's the tool? Cloud computing. For example, the first time IBM come up with a smart plan, a smart city, that's 2005 or 2006, around that time, there's no cloud computing yet, at the earliest emerging stage. And then we see what happens. And the smarter city and then gradually become IT infrastructure construction. But it's not DT, data technology. So they invested billions of dollars in the infrastructure level, and they collect so much data, but all the data become a burden to the government, to save, to archive the data or protect the data from hacking, right. Now, these days, if you have the cloud computing available, you can do real-time analytics to unleash the value in the first place, at the first moment you receive the data and then later on you know which data is more valuable, which data is of less value, and then you know how much you want to archive. >> Our Wikibon research team put out research this past year that said IT is no longer a department, it's everywhere, >> It's everywhere >> it supports your DT, data technology, it's a fabric. But one thing that's interesting going back to 2005 to now is not only the possibility for unlimited compute, is that now you're seeing wireless technologies significantly exploding in a good way, it's really happening. That's also going to be a catalyst for change. >> Definitely. >> What's your thoughts on how wireless connectivity, 'cause you have all these networks, you have to move data around, it has to be addressable, you have to manage security. That's a heavy load.\ what do you do, how are you guys doing that? >> Okay, very good question. We faced this challenge a couple of years ago, we realized that, because in Chinese domestic market, the users they are migrating from PC to mobile, and this create the mobile phone has wi-fi, right, so interacts with another AP, Access Point, right. So then how do we recognize our tracking, and recognizes ID identification, all this stuff, create huge headache to us, and this time, in this conference, we announce our solution for mobile, for mobile cloud. So what does that mean? So essentially, we have a cloud infrastructure product designed in order to do a real-time integration and do a data cleansing of the mobile data. I mean by mobile, and wireless as well. Wireless means even bluetooth, or even IoT, IoT solution also supported there. So this is a evolving process in the way. The first solution probably is less than perfect, but gradually, as we are expanding into more and more application scenario, and then we will amalgamate the solution and try to make it more robust. >> You guys have a good opportunity, and Alibaba Cloud certainly met with Karen Liu about the opportunity in North America and United States where I'm from. But Alibaba Cloud, and Alibaba Group, in the Alibaba Cloud has had a great opportunity, almost a green field, almost a clean sheet of paper, but you have a very demanding consumer base here in China. They're heavily on mobile as you pointed out, but they love applications. So the question I want to ask you is, and I'd love your thoughts on this. How do you bring that consumerization, its velocity, the acceleration of the changing landscape of the consumer expectation and their experience to small businesses and to enterprises? >> Ok, very good question. So user not just customer base, and the demanding customers in China trying to help us to harden our product, harden our solution, and to reduce the cost, the overall cost, and the economy of mass scale, economy of scale, and then once we reach that critical point, and then our service is inexpensive enough, and then the small and medium, SMB, small and medium business they could afford that. And in old days, SMB, they want to have access to high performance computing, but they do not have enough budget to afford the supercomputer. But these days now, because our product, our computation product, cloud product, big data product is efficient enough, so the total cost is affordable. And then you see that 80% of our customers of Alibaba, at least 80%, are actually SMB. So we believe the same practice can be applied to overseas market as well. >> You bring the best practices of the consumer and the scale of Alibaba Cloud to the small and medium-sized enterprises, and they buy as they grow. >> Exactly. >> They don't buy a lot upfront. >> Yeah, yeah, they buy on demand, as they need. >> That's the cloud, the benefit of the cloud. >> Exactly. >> Okay, the compute is great, you've got greatness with the compute power, it's going to create a renaissance of big data applications where you see that. What is your relationship with Intel and the ecosystem, because we see, you guys have the same playbook as a lot of successful companies in this open source era, you need horsepower and you need open source, what is Alibaba's strategy around the ecosystem, relationship with Intel, and how are you guys going to deal with partners? >> Yeah, first of all, so we really happy that we have Intel as our partner. In our most recent big data hackathon for the medical AI competition, and we just closed that competition, that data hackathon. Okay, very fascinating event, okay. Intel provided a lot of support. All the participants of this data hackathon, they do their computing leveraging on the Intel's products, because they do their image process. And then we provided the overall computing platform. Okay, this is a perfect example of how we collaborated with our technology partners. Beyond Intel, in terms of the ecosystem, first of all, we are open. We are building our ecosystem. We need partners. We need partners from pure technology perspective, and we also need partners from the traditional vertical sectors as well, because they provide us domain knowhow. Once we couple our cloud computing and big data technology with the domain knowhow, the subject matter expertise, well together the marriage will generate a huge value. >> That's fantastic, and believe me, open source is going to grow exponentially, and by 2025 we predict that it's going to look like a hockey stick. From the Linux foundation that's doing amazing work, you're seeing the Cloud Native Foundation. I want to get your thoughts on the future generation. >> Yeah, you mean open source? >> The future generation that's using open source, they're younger, you guys have tracked, you know the demographics in your employee base, you have a cloud native developer now emerging. They want to program the infrastructure as they go. They don't want to provision servers, they want the street lights to just work, whatever the project, the brains have to be in the infrastructure, but they want to be creative. You're bringing two cultures together. And you've got AI, it's a wonderful trend, machine learning is doing very well. How do you guys train the younger generation, what's your advice to people looking at Alibaba Cloud, that want to play with all the good toys? You got machine learning, you got AI, they don't want to necessarily baby, they don't want to program either. They don't want to configure switches. >> Yeah, very good question. Actually this is related to our product strategy. So in a way, like today we announce our ET Brain, so we are going to release this and share this as a platform to nurture all the creative mind, creative brains, okay, people, trying to leverage on this brain and then do the creative job, rather than worry about the underlying infrastructure, the basic stuff. So this is that part which we want to share with the young generation, tell them that unleash your creativity, unleash your imagination, don't worry about the hard coding part, and we already build the infrastructure, the backbone for you. And then image anything you think possible and then try to use ET Brain, try to explore that. And we provide the necessary tool and building blocks. >> And the APIs. >> And the APIs as well, yes. >> Okay, so I want to get your thoughts on something important to our audience, and that is machine learning, the gateway to AI. AI, what is AI? AI software, using cloud. Some will argue that AI hasn't really yet come on the scene but it's coming. We love AI, but machine learning is really where the action is right now, and they want to learn about how to get involved in machine learning. So what's your view on the role of machine learning, because now you have the opportunity for a new kind of software development, a lot of math involved, that's something that you know a lot about. So is there going to be more libraries? What's your vision on how machine learning moves from a bounded use case to more unbounded opportunities, because, I'm a developer, I want the horizontally scalable resource of the cloud, but I'm going to have domain expertise in a vertical application. So I need to have a little bit of specialism, and I want the scalability. So data's got to move this way and it's got to be up this way. >> Yes, yeah, okay, let me put it this way. So first off, for people who are really interested in AI, or they want to work on AI, my recommendation first of all, you got to learn some mathematics. Why, because all the AIs and machine learnings is talking about algorithms, and those algorithms are actually all about math, mathematics, the formula, and also the optimization, how to speed up the convergence of the algorithms, right. So all this maths is important, okay. And then if you have that math background, and then you have the capability to judge or to see next, which algorithm, or which machine software is suitable to solve the vertical problems. Very often the most popular algorithm may not be the right one to solve the specific vertical problems. So you're going to the way, capability to differentiate and to see that and make the right choice. That's the first recommendation. The second recommendation, try to do as many type of examples as possible, try to get your hands on, don't stop at looking at the function specification and oh, this is a function and input, output, da da da, but you need to get your hands dirty, get your hands on the real problem, the real data. So that you can have a feeling about how powerful it is or how bad or how good it is. Once you have this kind of experience, and then you do have capability, you gradually build up a cumulative capability to make a right choice. >> This is fascinating, Dr. Wanli, this is fantastic. I want to follow up on that because you're bringing up, in my mind I can almost see all these tools. There's an artisan culture coming on. You're seeing that. Dr. Wong discussed that with me yesterday. Artisans meeting technologists, scientists and creatives. UI, we're seeing evolutions in user experience that's more art. And so culture's important. But the machine learners of the algorithms, sometimes you have to have a lot of tools. If you have one tool, you shouldn't try to use tools for other jobs. So bring this up. How should a company who's architecting their business or their application look at tooling, because on one hand, there's the right tool for the right job, but you don't want to use a tool for a job that it's not designed for. To your point. Tools, what's your advice and philosophy on the kinds of toolings and when to engage platforms, relationship between platforms and tools. >> Okay, then put it this way. So, this is a decision based on a mixture of different criteria together. So first of all, from technology perspective, and secondly from the business perspective. From technology perspective I would say if your company's critical competence is technical stuff, and then you've got to have your own tool, your own version. If you only rely on some existing tool from other companies, your whole business actually is dependent on that, and this is the weakest link, the most dangerous link. But however, very often to develop your own version of the tool takes forever, and market wouldn't give you so much time. And then you need too strike a balance, how much I want to get involved for self development and how much for in-house development, and it's how much I want to buy in. >> And time. >> And time as well, yes. And another one is that you've got to look at the competitive landscape. If this tool actually has already existed for many years and many similar product in the market, and the problem is not a good idea to reproduce or reinvent, and then you're going to why not buy it, you take that for granted. And it think that's a fact, and then you build a new fact, right. So this is another in terms of the maturity of the tool, and then you need to strike a balance. And in the end, in the extreme case, if your business, your company is doing a extremely new, innovative, first of a kind study or service, you probably need some differentiate, and that differentiator probably is a new tool. >> Final question for you. For the audience in America, in Silicon Valley, what would you like to share from your personal perspective about Alibaba Cloud that they should know about? Or they might not know about and should know about. >> Okay, 'cause I worked in the US for 16 years. To be frank, I knew nothing about Alibaba until I came back. So as a Chinese overseas, I'm so ignorance about Alibaba until I came back. So I can predict, I can guess, more or less, in the overseas market, in US customers, they probably know not that much about Alibaba or Alibaba Cloud. So my advice and from my personal experience, I say, first of all, Alibaba is a global company, and Alibaba Cloud is a global company. We are going to go global. It's not only a Chinese company, for example. We are going to serve customers overseas market in Europe and North America and Southeast Asia. So we want to go global first. And second, we are not only doing the cloud. We are doing blending of cloud and big data and vertical solutions. I call this VIP. V for vertical, I for innovation. P for product. So VIP is our strategy. And the innovation is based upon our cloud product and big data product. >> And data's at the center of it. >> Data is the center of this, and we already got our data technique, our data practice from our own business, which is e-commerce. >> And you're solving some hard problems, the ET Brain's a great playground of AI opportunity. You must be super-excited. >> Yeah, yeah, right, right, okay. >> Are you having fun? >> Yes, a lot of fun. Very rewarding experience. A lot of dreams really come true. >> Well, certainly when you come to Silicon Valley, I know you have a San Mateo office, we're in Palo Alto, and this is theCUBE coverage of Alibaba Cloud. I'm John Furrier, co-founder of SiliconANGLE, Wikibon research and theCUBE, here in China covering the Alibaba Cloud, with Dr. Wanli, thanks for watching.

>> SiliconANGLE Media presents theCUBE! Covering AlibabaCloud's annual conference. Brought to you by Intel. Now, here's John Furrier... >> Hello everyone, I'm John Furrier, the co-founder of SiliconANGLE, Wikibon, and theCUBE, for our exclusive coverage in Hangzhou, China for Alibaba Cloud Conference here, it's a Cloud Computing Conference. The entire city is a cloud. We're here at the Intel booth with Peter Chen, who's the general manager of Products and Technology, for Data Center Group Sales of Intel Corporation. Peter, AI is the hottest topic, IoT, Alibaba Cloud, I mean, a huge event here mixing, kind of a cultural shift, generational shift, young developers. >> Definitely lots of crowd, you can see people surrounding us, right? So, artificial intelligence is definitely a hot word here in China for the past 12 months. Everybody's trying to figure it out, what's going on, how they can really use them, so we're very excited as well to really partner with Alibaba to really explore some of the potentials. >> I had a chance to speak with some of the Alibaba executives, and obviously, a strategic partnership with Intel, pretty strategic, and it matters what's inside the Cloud. But it's not an Intel inside like a PC. The AI is showing that there's a little bit of Intel on everything, from IoT, industrial IoT to data center. It's a range of technology that's powering a new kind of software. This is where AI is shining. We're seeing that with machine learning and as data driven technology. So, I got to ask you. What is the view from Intel on AI? Obviously, we see the commercials, we see the technology from Intel. How does that translate to your view on AI? What's that view? >> So, essentially today's AI, artificial intelligence, is powered by three factors, the amount of data, the new algorithms, and lastly the compute power. And Intel has historically been the leader of create and compute. So, for the past many years, we has always been generating new compute powers into the cloud and data centers as well as PCs. But going forward as we look at applying AI to different usages like autonomous driving, for example, you cannot expect everything to be done just in the cloud because we need the real data to be inputted from a car, for instance, all the cameras, all the sensors. So, we do definitely see a need of actually faster processors at the edge as well to constantly bring in the data back into the cloud, so they have an autonomous feedback loop, make sure there will be right decision making. >> Yeah, so Cloud drives this, right? So, it's not just Cloud though, it's software. There's exponential growth in open source software that's causing a Renaissance in the developer community. You're seeing it here in China, a lot of young demographics here. Software and data's tsunami going on. You need compute power. >> Yes, yes. I think, everybody knows Intel is a hardware company, but we do have a very large effort on engaging a software ecosystem. From the old days on engaging Linux, the cloud different software stack, and working with CSPs like Alibaba in China to really make sure they can create and write the new latest software AI framework and taking the most advantage of our hardware platform as well. So, that's something that we've been very focusing on. >> And one of the themes here is the IoT for traffic in China. Obviously, if you've been here, you know it's kind of congested. But Alibaba is giving a lot of talks on how they're using data in this cloud city for traffic, which is an example of IoT, Internet of Things, but applied to the real world. That's where the AI kind of connects with the data. Is that kind of where it's going? >> Yeah, so I think this is a great application, as you just mentioned. And Alibaba calls it City Brain. So, essentially, imagine a normal city like in China, can easily go five million, 10 million people. The amount of people and the amount of traffic that goes on the road every day. So, if the city is able to utilize all these videos' stream of data, feedback from different traffic intersections, and be able to direct traffics and control the traffic lights dynamically, using artificial intelligence, you'd actually solve a lot of the city's congestion problem. So, I think this is where we are seeing a lot of application being explored in China, they're using very innovative, different ways by Alibaba. >> Peter, I've got to ask you because one of the things we're seeing in theCUBE and Wikibon Research is the growth of new kinds of ecosystems. Karen Liu, who runs the America's, general manages for America's Alibaba, said to me that ecosystem is super important for Alibaba as an example. But a new kind of ecosystem is developing. Cloud service providers are becoming a new hot growth area because the specialty of building applications in the cloud is not like it was kind of in the old days. You got to have a little bit of a cloud native mindset, but yet, domain expertise, whether it's traffic or a certain vertical solution. So, it's a little bit of both. Always often scalable, yet specialism. This is going to create a lot of opportunities for cloud service providers. What's your view on that from Intel's perspective? How are you guys seeing that market? Do you agree? And what are you guys looking at, at that market? >> So, obviously cloud service provider, the likes of Alibaba or Amazon, are one of our fastest growing customer base over the past five years. And in the near future, we expect this trend to continue to grow. We definitely see CSPs as a leading edge of driving innovation because they are not just the leading edge of driving consumer usages but they also, like the City Brain project, they've been really close on solving the enterprise problem as well with public cloud. So, I think we're very excited to have the opportunity to be a close partner with a CSP like Alibaba to really help them, providing our latest hardware technology to allow them to drive innovation on top of this offer and with the programs and the algorithms. >> How are they, how are those big cloud service providers, or CSPs like Alibaba, they're a big one, they're the fourth cloud in the world, enabling their CSPs? Because I was just talking to someone on the floor here, an ISV in the old world, who was telling me that he's now a cloud service provider, so you have now this nice balance in the ecosystem developing. You guys see the same thing? How do you guys, looking at that? >> So, this is what we call a hybrid situation. So, while the big CSPs like Alibaba, they have a lot of competency and they have a lot of internal engineering, it may not make sense for them to create every single application in the world. So, there may be some legacy enterprise application, for instance, a CRM software in China, maybe it was really popular, for them to forge a collaboration with the leading company Alibaba to translate their on-prem software stack into a cloud solution. So, I think we definitely see a lot of that collaboration happening, to take the best of the best from the legacy as well as the new public cloud environment to really make the better service for the companies and the customers. >> Create ecosystem opportunity. Okay, so I got to ask. What is the Intel relationship that you guys are doing on your end with Alibaba Cloud? Obviously, they're taking names, they're kicking butt. They're doing well. They're going global. They're not just in China. They're the first cloud provider here in China to go outside the mainland. Obviously, they're in the US, they're in Silicon Valley, our backyard. What's the collaboration? Share the relationship. >> We work very closely with Alibaba. Like you said, they're now the leading cloud service provider in China. They're starting going abroad. And we as an ingredient, knowledge provider perspective, we have a very close collaboration with them, sharing with them our technologies on hardware roadmap as well as software enabling to make sure they can take full advantage of it. So, we're very excited to see the growth of Alibaba over the past few years, and we look forward to seeing them continue to expand their business together with us. >> Yeah, great company. So, I got to ask you, one of the collaborations that got my attention was the, I don't want to say hack-a-thon, it was a competition, it was the AI competition called Tianchi that you guys were a part of with Alibaba. What was Intel's role in that? I saw some of the winners earlier. I didn't get a chance to get the specifics, but take me through this AI competition that Alibaba did with these entrepreneurs. >> So, I'm very, actually very excited. I just talked to one of the winning team just now. So, what happened is, when we talk about artificial intelligence, today it's a lot about image recognition, voice recognition but that's just pure technology. So, what Alibaba decide to do, which in terms of partner, is we created a medical image contest. So, we pick a particular subject, for instance, lung cancer and we invited 16 local hospitals to provide some of the image data of the patients anonymously, and then we opened it up for the software ecosystem, the academia, professors, the schools, and say, hey, why don't you come in and try to compete on the image recognition accuracy based on those X-ray images, using these images? So, it takes about six, we have overwhelming turnout. We have about 3,000 teams from 20 different countries applying to join in the contest, and then we just select the winner yesterday. So, basically, the three winning teams, two of them are from the best universities here in China, one of them is from overseas. And again, Intel's role in this is we provide a lot of consultation help. First of all, we provide the hardware system based on our Xeon Phi clusters, and on top of that, we provide a lot of the software tools, Caffe, image recognition libraries, Intel material libraries to really help the contestant to be able to use the Intel hardware for the maximum to drive the best performance. >> And so, you guys provided the technology, Alibaba the Cloud, and let these guys just take. What was the results? Was there any success? Was there a winner? >> There was a winner. I think the big winner was Beijing University. But I think overall, we are not just excited just because of this specific winners but really the larger intent. If you can imagine in a country like China, there's a lot of people, meaning there's a lot of patients at different part of the country, and not every tier two, tier three city have the same resource or access of the best doctors. If we're able to simplify the lung cancer image recognition to be able to provide this as a tool for all the tier two, tier three cities of China, imagine how much this will change. >> It's a societal impact. >> Definitely. >> And you've got a collective intelligence. It's almost like an open source kind of thing, where the more people doing it. >> It gets better, it gets better. >> The fly wheel. >> And then, we have definitely a lot of hospitals who want to really take advantage of this as well. So, we're really glad on the results of this first round, and I think Alibaba will do a next round with a different subject as well, and we're looking forward to partnering with them again. >> That's inspirational. Okay, great to have you on. Thanks for the commentary. Exclusive coverage. Final thought, what's your thoughts on the event? Where's AI going? Where do you see this trajectory of Alibaba and Intel going? >> So, definitely the event's wonderful and great. This is my third year here. It gets just bigger and bigger every time. So, I'm looking forward to come back for the next couple of years again. Our collaboration with Alibaba has been very close. We work with each other deeply, with our engineers' collaboration, and I look forward to continuing to bring out more successful projects. >> And they're really bringing together, not just science and developers, but artists. You've got a music festival here, feels like South by Southwest meets a developer conference. Societal impact, traffic, solving problems, lung cancer, big data, and data is changing the world. Now, you need the compute power, you need the analytics. Of course, you need SiliconANGLE and theCUBE and Wikibon, exclusive coverage here in China of the Alibaba Cloud Conference. Thanks for watching.