(futuristic electronic music) >> Live from Las Vegas, its theCUBE covering AWS re:Invent 2018. Brought to you by Amazon Web Services, Intel, and their ecosystem partners. >> Hey welcome back everyone. Live here in Las Vegas with AWS, Amazon Web Services, re:Invent 2018's CUBE coverage. Two sets, wall-to-wall coverage here on the ground floor. I'm here with Dave Vellante. Dave, six years we've been coming to re:Invent. Every year except for the first year. What a progression. We got great news. Always raising the bar, as they say at Amazon. This year, big announcements. One of them is blockchain. Really kind of laying out early formation of how they're going to roll out, thinking about blockchain. We're here to talk about here, with Rahul Pathak, who's the GM of analytics, and data lakes, and blockchain. Managing that. And Shawn Bice who's the vice president of non-relational databases. Guys, welcome to theCUBE. >> Thank you. >> Thank you, it's great to be here. >> I wish my voice was a little bit stronger. I love this segment. You know, we've been doing blockchain. We've been following one of the big events in the industry. If you separate out the whole token ICO scam situation, token economics is actually a great business model opportunity. Blockchain is an infrastructure, a decentralized infrastructure, that's great. But it's early. Day one really for you guys in a literal sense. How are you guys doing blockchain? Take a minute to explain the announcement because there are use cases, low-hanging use cases, that look a lot like IoT and supply chain that people are interested in. So take a minute to explain the announcements and what it means. >> Absolutely, so when we began looking at blockchain and blockchain use cases, we really realized there are two things that customers are trying to do. One case is really keep an immutable record of transactions and in a scenario where centralized trust is okay. And for that we have Amazon QLDB, which is an immutable cryptographically verifiable ledger. And then in scenarios where customers really wanted the decentralized trust and the smart contracts, that's where blockchain frameworks like Hyperledger Fabric and Ethereum play a role. But they're just super complicated to use and that's why we built Managed Blockchain, to make it easy to stand up, scale, and monitor these networks, so customers can focus on building applications. And in terms of use cases on the decentralized side, it's really quite diverse. I mean, we've got a customer, Guardian Life Insurance, so they're looking at Managed Blockchain 'cause they have this distributed network of partners, providers, patients, and customers, and they want to provide decentralized verifiable records of what's taking place. And it's just a broad set of use cases. >> And then we saw in the video this morning, I think it was Indonesian farmers, right? Wasn't that before the keynote? Did you see that? It was good. >> I missed that one. >> Yeah, so they don't have bank accounts. >> Oh, got it. >> And they got a reward system, so they're using the blockchain to reward farmers to participate. >> So a lot of people ask the question is, why do I need blockchain? Why don't you just put in database? So there are unique, which is true by the way, 'cause latency's an issue. (chuckles) Certainly, you might want to avoid blockchain in the short term, until that gets fixed. Assume that every one will get fixed over time, but what are some of the use cases where blockchain actually is relevant? Can you be specific because that's really people starting to make their selection criteria on. Look, I still use a database. I'm going to have all kinds of token and models around, but in a database. Where is the blockchain specifically resonating right now? >> I'll take a shot at this or we can do it together, but when you think of QLDB, it's not that customers are asking us for a ledger database. What they were really saying is, hey, we'd like to have this complete immutable, cryptographically verifiable trail of data. And it wasn't necessarily a blockchain conversation, wasn't necessarily a database conversation, it was like, I really would like to have this complete cyrptographic verifiable trail of data. And it turns out, as you sort of look at the use cases, in particular, the centralized trust scenario, QLDB does exactly that. It's not about decentralized trust. It's really about simply being able to have a database that when you write to that database, you write a transaction to the database, you can't change it. You know, a typical database people are like, well, hey, wait a second, what does immutable really mean? And once you get people to understand that once that transaction is written to a journal, it cannot be changed at all and attached, then all of a sudden there's that breakthrough moment of it being immutable and having that cryptographic trail. >> And the advantage relative to a distributive blockchain is performance, scale, and all the challenges that people always say. >> Yeah, exactly. Like with QLDB, you can find it's going to be two to three times faster cause you're not doing that distributing consensus. >> How about data lakes? Let's talk about data lakes. What problem were you guys trying to solve with the data lakes? There's a lot of them, but. (chuckles) >> That's a great question. So, essentially it's been hard for customers to set up data lakes 'cause you have to figure out where to get data from, you have to land it in S3, you've got to secure it, you've then got to secure every analytic service that you've got, you might have to clean your data. So with lake formation, what we're trying to do is make it super easy to set up data lakes. So we have blueprints for common databases and data sources. We bring that data into an S3 data lake and we've created a central catalog for that data where customers can define granular access policies with the table, and the column, and the row level. We've also got ML-based data cleansing and data deduplication. And so now customers can just use lake formation, set up data lakes, curate their data, protect it in a single place, and have those policies that enforce across all of the analytic services that they might use. >> So does it help solve the data swamp problem, get more value out of the data lake? And if so, how? >> Absolutely, so the way it does that is by automatically cataloging all datas that comes in. So we can recognize what the data is and then we allow customers to add business metadata to that so they can tag this as customer data, or PII data, or this is my table of sales history. And that then becomes searchable. So we automatically generate a catalog as data comes in and that addresses the, what do I have in my data lake problem. >> Okay, so-- >> Go ahead. >> So, Rahul, you're the general manager. Shawn, what's your job, what do you do? >> So our team builds all the non-relational databases at Amazon. So DynamoDB, Neptune, ElastiCache, Timestream, which you'll hear about today, QLDB, et cetera. So all those things-- >> Beanstalk too, Elastic Beanstalk? >> No we do not build Beanstalk. >> Okay, we're a customer of DynamoDB, by the way. >> Great! >> We're happy customers. >> That's great! >> And we use ElastiCache, right? >> Yup, the elastic >> There you go! >> surge still has it. >> So-- >> Haven't used Neptune yet. >> What's the biggest problem stigmas that you guys are trying to raise the bar on? What's the key focus as you get this new worlds and use cases coming together? These are new use cases. How are you guys evaluating it? How are you guys raising the bar? >> You know, that's a really good question you ask. What I've found in my experience is developers that have been building apps for a long time, most people are familiar with relational databases. For years we've been building apps in that context, but when you kind of look at how people are building apps today, it's very different than how they did in the past. Today developer do what they do best. They take an application, a big application, break it down into smaller parts, and they pick the right tool for the right job. >> I think the game developer mark is going to be a canary in the coal mine for developers, and it's a good spot for data formation in these kind of unstructured, non-relational scenarios. Okay, now all this engagement data, could be first person shooter, whatever it is, just throw it, I need to throw it somewhere, and I'll get to it and let it be ready to be worked on by analytics. >> Well, yeah, if you think about that gamer scenario, think about if you and I are building a game, who knows if there's going to be one user, ten players, or 10 million, or 100 million. And if we had 100 million, it's all about the performance being steady. At 100 million or ten. >> You need a fleet of servers. (John laughing) >> And a fleet of servers! >> Have you guys played Fortnite? Or do you have kids that play? >> I look over my kid's shoulder. I might play it. >> I've played, but-- >> They run all their analytics on us. They've got about 14 petabytes in S3 using S3 as their data lake, with EMR and Athena for analytics. >> We got a season-- >> I mean, think about that F1 example on keynotes today. Great example of insights. We apply that kind of concept to Fortnite, by the way, Fortnite has theCUBE in there. It's always a popular term. We noticed that, the hastag, #wherestheCUBEtoday. (Rahul chuckling) I couldn't resist. But the analytics you could get out of all that data, every interaction, all that gesture data. I mean, what are some of the things they're doing? Can you share how they're using the new tech to scale up and get these insights? >> Yeah, absolutely. So they're doing a bunch of things. I mean, one is just the health of the systems when you've got hundreds of millions of players. You need to know if you're up and it's working. The second is around engagement. What games, what collection of people work well together. And then it's what incentives they create in the game, what power ups people buy that lead to continued engagement, 'cause that defines success over the long term. What gets people coming back? And then they have an offline analytics process where they're looking at reporting, and history, and telemetry, so it's very comprehensive. So you're exactly right about gaming and analytics being a huge consumer of databases. >> Now, Shawn, didn't you guys have hard news today on DynamoDB, or? >> Yeah today we announce DynamoDB On-Demand, so customers that basically have workloads that could spike up and then all of a sudden drop off, a lot of these customers basically don't even want to think about capacity planning. They don't want to guess. They just want to basically pay only for what they're using. So we announced DynamoDB On-Demand. The developer experience is simple. You create a table and you putyour read/write capacity in the on-demand mode, and you literally only pay for the request that your workload puts through system. >> It's a great service actually. Again, making life easier for customers. Lower the bill, manage capacity, make things go better, faster, enables value. >> It's all about improving the customer experience. >> Alright, guys, I really appreciate you coming in. I'm really interested in following what you guys do in the future. I'm sure a lot of people watching will be as well, as analytics and AI become a real part of, as you guys move the stack and create that API model for, what you did for infrastructure, for apps. A total game changer, we believe. We're interested in following you guys, I'm sure others are. Where are you going to be this year? What's your focus? Where can people find out more besides going to Amazon site? Is there certain events you're going to be at? How do people get more information and what's the plans? >> There's actually some sessions on lake formation, blockchain that we're doing here. We'll have a continuous stream of summits, so as the AWS Summit calendar for 2019 gets published that's a great place to go for more information. And then just engage with us either on social media or through the web and we'll be happy to follow up. >> Alright, well, we'll do a good job on amplifying. A lot of people are interested, certainly blockchain, super hot. But people want better, stronger, more stable, but they want the decentralized immutable database model. >> Cryptographically verifiable! >> And see as everyone knows. >> Scalable! >> Anyone who wants to keep those, they talk about CUBE coins but I haven't said CUBE coin once on this episode. Wait for those tokens to be released soon. More coverage after this short break, stay with us. I'm John Furrier, and Dave Vellante, we'll be right back. (futuristic buzzing) (futuristic electronic music)

>> Live from Las Vegas, it's theCUBE! Covering AWS re:Invent 2018. Brought to you by Amazon web services, Intel, and their ecosystem partners. >> Hey welcome back everyone, here at AWS re:Invent 2018, their sixth year of theCUBE coverage, two sets wall-to-wall coverage here, two more sets in other locations, getting all the content, bringing it in, ingesting it into our video cloud service on AWS, ah, Dave, >> Lot of content, John. >> Lot of people don't know that we have that video cloud service, but we're going to have a lot of fun, ton of content, ton of stories, and a special analyst segment, Jerry Chen, guest here today, CUBE alumni, famous Venture Capitalist and Greylock partners, partnering with Reid Hoffman, the founder of LinkedIn, great set of partners at Greylock , great firm, tier one, doing a lot of great deals, Rockset, recent one. >> Thanks, yeah. >> You're also, on the record, these six years ago, calling the shot of Babe Ruth predicting the future. You've got a good handle on, you've got VM where you have the cloud business, now you're making investments, you're seeing a lot of stuff on the landscape, certainly, as a Venture Capitalist, you're funding projects, what better time now of innovation to actually put money to work, to hit market share, and then the big guys are getting bigger, they're creating more robust platforms, game is changing big-time, want to get your perspective, Dave, so, Jerry, what's your take on the announcements, slew of announcements, which ones jumped out at you? >> I think there's kind of two or three areas, there's definitely the hybrid cloud story with the Outpost, there's a bunch of stuff around ML and AI services, and a bunch of stuff on data and storage, and for me I think what they're doing around the ML services, the prediction, the personalization, the text OCR, what Amazon's doing at that app layer is now creating AI building blocks for modern application, so you want to do forecasts, you want to do personalization, you want to do text analysis, you have a simple API to basically build these modern apowered apps, he's doing to the app infrastructure layer what he's done to the cloud infrastructure layer, by deconstructing these services. >> And API is also the center, that's what web services are, so question for you is, do you see that the core cloud players, Aussie, Amazon, Bigly, Google, Microsoft, others, it's a winner take most, you called that six years ago, and that's true, but as they grow there's going to be now a new cloudification going on for business apps, new entrepreneurs coming to market, who's vulnerable, who wins, who loses, as this evolution continues because it's going to enable a lot of opportunity. >> Yeah, well I mean Amazon in cloud in general is going to create a lot of winners and losers, like you said, so I think you have a shift of dollars from on prem and old legacy vendors, databay storage, compute, to the cloud, so I think there's a shift of dollars, there are winner and losers, but I think what's going to happen is, with all these services around AI, ML, and Cloud as a distribution model, a lot of applications are going to be rebuilt. So I think that the entire application stack from all the big SaaS players to small SaaS companies, you're going to see this kind of a long tale of new SaaS applications being built on top of the Cloud that you didn't see in the past. >> And the ability to get to markets faster, so the question I have for you is, if you're an entrepreneur out there, looking for funding and I can to market quicker, what's the playbook, and two, Jassie talked on stage about a new persona, a new kind of developer, one that can rethink and reimagine and reinvent something that someone else has already done, so if you're an entrepreneur, you got to think to take someone else's territory, so how does an entrepreneur go out and identify whose lunch to eat, so if I want to take down a company, I got to have a strategy, how do I use the cloud to >> I think it's always a combination when a founder in a thing attacks your market it's a combination of where are the dollars, where can I create some advantage IP or advantage angle, and thirdly where do I have a distribution advantage, how can I actually get my product in the hands of the users differently? And so I think those are the three things, you find intersection of a great market, you have a unique angle, and you have a unique route to market, then you have a powerful story. So, you think about cloud changing the game, think about the mobile app you can consist of, for consumers, that is also a new platform, a new distribution method, the mobile app stores, and so what happened, you had a new category of developers, mode developers, creating this long tale, a thousand thousand apps, for everything from groceries to cars to your Fantasy Football score. So I think you're going to see distribution in the cloud, making it easy to get your apps out there, going to see a bunch of new markets open up, because we're seeing verticals like healthcare, construction, financial services, that didn't have special apps beforehand, be disrupted with technology. Autodesk just bought PlanGrid for 800 million dollars, I mean that's unheard of, construction software company. So you can see a bunch of new inverdics like that be opened up, and then I think with this cloud technology, with compute storage network becomes free and you have this AI layer on top of it, you can power these new applications using AI, that I think is pretty damn exciting. >> Yes, you described this sort of, we went from client server to the cloud, brought a whole bunch of new app providers, obviously Salesforce was there but Workday, Service Now, what you described is a set of composeable digital services running on top of a cloud, so that's ripe for disruption, so do I have to own my own data centers if I'm big SaaS company, what happens to those big guys? >> I don't think you have to, well, you don't have to own your own data center as a company, but you could if you wanted to, right, so at some point in scale, a lot of big players build their own data centers, like AirBNB is on Amazon, but Dropbox built their own storage on Amazon early, then their own data center later. Uber has their own data center, right, so you can argue that at some point of scale it makes sense to build your own, so you don't need to be on Amazon or Google as your start, but it does give you a head start. Now the question is, in the future, can you build a SaaS application entirely on Amazon, Azure, or Google, without any custom code, right, can you hide read write call private SaaS, like a single instance of my SaaS application for you, John, or for you, Dave, that's your data, your workflow, your information personalized for you, so instead of this multi-tenet CRM system like Salesforce, I have a custom CRM system just for Dave, just for Jeff, just for Jerry, just for theCUBE, right? >> I think yes, for that, I think that's definitely a trend I would see happening. >> It's what Infor is trying to do, right, Charles Phillips says "Friends don't let friends "build data centers," but they've still got a big loss in legacy there, but it's an interesting model, focused on verticals or microverticals or like the healthcare example that you're giving, and lot of potential for something. >> Well here's why I think I like this because, I think, and I said this before in theCUBE maybe it's not the best way to say it is that, if you look at the benefit of AI, data-driven, the quality of the data and the power of the compute has to be there. AI will work well with all that stuff, but it's also specialized around the application's use case. So you have specialism around the application, but you don't have to build a full stack to do that, you could use a horizontally scalable cloud distribution system in your word, and then only create custom unique workloads for the app, where machine learning's involved, and AI, that's unique to the app, that's differentiation, that could be the business model, or the utility. So, multitenancy could exist in theory, at the scalable level, but unique at the top of the level so yes I would say I'd want that hosted in the most customized, agile, flexible way. So I would argue that that's the scenario. >> I think that's the future, I mean one of my, I think you were saying, Dave, friends don't let friends build data centers anymore, it's you probably don't need to build a data center anymore because you can actually build your own application on top of one of the two or three large cloud providers. So it's interesting to see what happens the next three, four years, we're going to see kind of a thousand flowers bloom of different apps, not everyone's going to make it, not everyone's going to be a huge Salesforce-like outcome, but there'll be a bunch of applications out there. >> And the IoT stuff is interesting to me, so observing a lot of what the IT guys are doing, it reminds me of people trying to make the Windows mobile phone, they're just trying to force IT standards down the IoT, what I've seen from AWS today is more of a bottoms up approach, build applications for operations technology people, which I think is the right way to go, what do you see in an IoT, IoT apps, what's the formula there? >> I think what Amazon announced today with their time series database, right, their Timestream prediction engine, plus their Outpost offering with the Vmware themselves, you're really seeing a combination of IoT and Edge, right, it's the whole idea is, one, there's a bunch of use cases for time series in IoT, because sentry data, cameras, self-driving cars, drones, et cetera, there's more data coming at you, it adds all of that. >> And Splunk has proven that big-time. >> Correct, Splunk's let 18 billion Marcap company, all on time series data, but number two, what's happening is, it's not necessarily centralized data, right, it's happening at the edge, your self-driving car, your cell phone, et cetera, so Outpost is really a way for Amazon to get closer to the edge, by pushing their compute towards your data center, towards remote office, branch office, and get closer to where the data is, so I think that'll be super interesting. >> Well the Elastic Inference engine is critical, now we got elasticity around inference, and then they got the chip set that worked Inferentia, that can work with the elastic service. That's a powerful combination. >> The AI plumbing war between Google and TetraFlow as technology there's like PyTorch, Google TPUs versus what Amazon is doing with inference chips today, versus what I'm sure Microsoft and else is doing, is fascinating to watch in terms of how you had a kind of a Intel Nvidia duopoly for a long time, and now you have Intel, Nvidia, and then everyone from Amazon, Google, Microsoft doing their own soul again, it's pretty fascinating to watch. >> What was the stat, he said 85% of the TensorFlow, cloud TensorFlow's running on AWS? >> Makes a lot of sense, I think he said Aurora's customers logoslide doubled, but let's break down real quick, to end the segment with the key areas that we see going on, at least my perspective, I want to get your reaction. Storage, major disruption, he emphasized a lot of that in the keynote, spent a lot of time on stores, actually I think more than EC2 if you look at it, two, databases, database war, storage rate configuration, and a holy trinity of networking, storage, and compute, that's evolving, databases, SageMaker, machine learning. All there and then over the top, yesterday's announcement of satellite as a service, that essentially kills the edge of the network, cause there is no edge if we have space satellites shooting connectivity to any device the world is now, there's no more edge, it's everywhere. So, your thoughts, those areas. Which one pops out as the most surprising or most relevant? >> I think it's consistent Amazon strategy, on the lowest layer they're trying to draw the cost to zero, so on storage, cheaper cheaper cheaper, they're driving the bottom layer to zero to get all your data. I think the second thing, the database layer, it makes sense, it's not open-source, right, time scale or time series, it's not, Timestream's not their open-source database, it's their own, so open-source, low cost, the lowest layer, their database stuff is mostly their own, Aurora, Dynamo, Timestream, right, because there's some level lock in there, which I think customers are worried about, so that's clever, it's not by accident, that's all proprietary, and then ML Services, on top of that, that's all cares with developers, and it's API locking, so clearly low-cost open-source for the bottom, proprietary data services that they're trying to own, and then API's on top of it. And so the higher up in the stack, the more and more Amazon, you look, the more and more Amazon you have to adopt as kind of a lock in stack, so it's a brilliant strategy the guys have been executing for the past six, seven years as you guys have seen firsthand, I think the most exciting thing, and the most shocking thing to me is this move towards this battle for the AI front, this ML AI front, I think we saw ML's the new sequel, right, that's the new war, right, against Amazon, Google, and Microsoft. >> And that's the future of applications, cause this is >> But you're right on, it's a knife fight for the data, and then you layer on machine intelligence on top of that, and you get cloud scale, and that's the innovation engine for the next 10 years. >> Alright Jerry Chen just unpacked the State of the Union of cloud, of course as an investor I got to ask the final question, how are you investing to take advantage of this wave, versus being on the wrong side of history? >> I have framers for everything, there's a framer on how to attack the cloud vendors, and so I'm looking at a couple things, one, a seams in between the clouds, right, or in between services, because they can't do everything well, and there were kind of these large continents, Amazon, Google, Azure, so I'm looking for seams between the three of them, I'm looking for two, deep areas of IP that they're not going into that you actually have proprietary tap, and then verticals of data, like source of the data, or workflows that these guys aren't great, and then finally kind of cross-data cross-cloud solution, so, something that gives you the ability to run on prem, off prem, Microsoft, Google, Azure. >> Yeah, fill in the white spaces, there are big white spaces, and then hope that could develop into, good. Jerry Chen, partner in Greylock , partners formerly Vmware part of the V Mafia, friend of theCUBE, great guest analysis here, with Dave Vellante and John Furrier, thanks for watching us, stay with us, more live coverage, day two of three days of wall-to-wall coverage at re:Invent, 52,000 people, the whole industry's here, you can see the formations, we're getting all of the data, we're bringing it to you, stay with us.