>> live from San Francisco, celebrating 10 years of high tech coverage. It's the Cube covering Veum World 2019 brought to you by IBM Wear and its ecosystem partners. >> Welcome back to the Cube. It's the Emerald 2019 our 10th year water wall coverage. Three days, two sets, lots of content. Instrument of my co host is Justin Warren. And one of the big stories coming into the show is VM Wear actually went on an acquisition spree. A hold number of acquisitions. Boston based Carbon Black over $2 billion Pivotal brought back into the fold for also, you know, around that ballpark of money on Happy to Welcome to the program. One of those acquisitions, such and Conti, is sitting to my right. He's the co founder of Hana is also a professor at Stanford University. Thank you so much for joining us and joining us. Also for the segment. Shakeri Air, the executive vice president general manager of Telco Edge Cloud at VM Wear, Shaker said, Yes, there's a lot of acquisitions not to play favorites, but maybe this is his favorite. No question about it. All right. Eso such in, you know, boy, you know the Paolo Alto Stanford connection. We were thinking back, You know, the Founders Of'em where, of course, you know came from Stanford. Many acquisitions over the year, including the mega next era acquisition. You know, quite a few years ago, I came out of Stanford. Give us what was the genesis in the Why of Hana. >> It's actually interesting Stanford Connection to So I've been a faculty at Stanford for the last 10 years on dhe. I have seen the SD and moment very close on up front on one of the dirty secrets of S. T M says it makes the netbooks programmable, but someone still has to write the programs on. So that's usually a very complex task on the pieces beyond the company was, Can we use the eye to learn how to program the network rather than humans having to program the network to do management or optimization? So the division really waas can be built? A network that learned how to optimize itself learns how to manage itself on the technology we're building. Is this a pipeline that basically tries to deliver on that for mobile? >> It's great, Sachin, you know, my background is networking and it feels like forever. We've been hooking well. We need to get people from the cli over to the gooey. But we know in today's rightly complex world, whether it's a I or just automation, humans will not be able to keep up with it. And, you know, we know that that's where a lot of the errors would happen is when we entered humans into doing some of this. So what are some of the key drivers that make this solution possible today that, you know, it might not have been able to do done when when one train was first rolling out the first S t n? >> Yeah, talk about it in three dimensions. The one is, Why do we need it today? Right on. Then what is being what is happening that is enabling this today, right? So, apart from what I talked about Stu and I think the other big driver is, the way I like to think about it is that the Internet is going from a means of consumption to a means of control and interaction. So, increasingly, the application to BC driving the next big decade, our very way of controlling things remotely or the network like a self driving car, or be in interacting but very highly rich visual content like E. R. India. So the applications are becoming a lot more demanding on the Net. At the same time, the network is going through a phase off, opening up on becoming disaggregated network complexity is increasing significantly. So the motivation behind the company and why I thought that was the right time to start the company was these two friends are gonna collide with five coming along the applications that are driving five g and then at the complexity increasing our five. So that's why we started the company. What actually is enabling. This is the fact that we have seen a lot of progress with the eye over the last few years. It hasn't really. It hadn't really been applied at scale to networks and specifically mobile that book. So we definitely saw no, actually there, but increasingly, ah, lot of the infrastructure that is being deployed there was more and more telemetry available. There was more and more data becoming available and that also obviously feet this whole engine. So I think the availability of all of these Big Data Technologies Maur data coming in from the network and the need because of these applications and that complexity. I think there's a perfect confluence >> that there's lots of lots of II floating around at the moment, and there's different flavors of it as well. So this machine learning there's Aye aye, sir. When when you say that there's there's a I behind this What? What particular kind of machine learning or a Y you're using to drive these networks? >> This a few different techniques because the problems we solve our anomaly detection off. Then problems are happening in the network predicting how network conditions are going to evolve. For example, predicting what your devices throughput is gonna be the next 30 seconds. We're also learning how to control the knobs in the neck using AI ai techniques. So each of these has different classes of the eye techniques. So, for example, for control we're using reinforcement learning, which is the same technique that Google used to kind of been on alphago. How do you learn how to play a game basically, but area the game you're playing it optimizing the network. But for the others, it's a record of neural networks to do predictions on Time series data. So I think it's a combination of techniques I wouldn't get to wherever the techniques. It's ultimately. But what is the problem you're trying to solve? And then they picked the right technique to solve it, >> and so on that because the aye aye is actually kind of stupid in that it doesn't know what they wouldn't. What an optimized network looks like. We have to show it what that is. So what? How do you actually train these systems to understand? But what is an optimized network? What? How does how does that tell you? Define this is what my network optimal state should be. >> So that's a great question, because in networking like that, any other discipline that wants to use the eye. There's not a lot of label data. What is the state I want to end up at what is a problem state or what is a good state? All of this is labels that someone has to enter, and that's not available axe kid, and we're never gonna be able to get it at the scale we wanted. So one of our secret sauce is if you will, is semi supervised learning but basic ideas that we're taking a lot of domain knowledge on using that domain knowledge to figure out what should be the right features for these models so that we can actually train these models in a scalable fashion. If you just throw it a lot of data any I model, it just does not converge. Hardly constructive features on the other thing is, how do I actually define what are good kind of end state conditions? What's a good network? And that's coming from domain knowledge to That's how we're making I scale for the stomach. >> I mean, overall, I would say, as you look at that, some of the parameters in terms of what you want to achieve are actually quite obvious things like fewer dropped calls for a cellular network. You know, that's good. So figuring out what the metrics need to be and what the tuning needs before the network, that's where Hana comes in in terms of the right people. >> All right, so shake her. Give us a little bit of an understanding as to where this fits into the networking portfolio. You know, we heard no we heard from Patty or two ago. You know what would have strong push? Networking is on the NSX number. Speaks for itself is what's happening with that portfolio? >> No, absolutely. In fact, what we're doing here is actually broader than networking. It's sort off very pertinent to the network off a carrier. But that is a bulk off their business, if you will. I think if you sort of go back and look at the emirs of any any, any vision, this is the notion of having any cloud in any application land on any cloud and then any device connected to those applications on that any cloud side we are looking at particularly to cloud pools, one which we call the Telco cloud and the other is the edge cloud. And both of these fortuitously are now becoming sort of transforming the context of five G. So in one case, in the telco cloudy or looking at their core and access networks, the radio networks, all of this getting more cloud ified, which essentially leads toe greater agility in service deployment, and then the edge is a much more distributed architecture. Many points over which you can have compute storage network management and security deployed. So if you now think about the sort of thousands off nodes on dhe virtualized clouds, it is just impossible to manage this manual. So what you do need is greater. I mean, orders of magnitude, greater automation in the ability to go and manage and infrastructure like this. So, with our technology now enhanced by Johanna in that network portfolio in the Telco Edge Cloud portfolio, were able to go back to the carriers and tell them, Look, we're not just foundational infrastructure providers. We can also then help you automate help you get visibility into your networks and just help you overall manager networks better for better customer expedience and better performance. >> So what are some of the use coasters that you see is being enabled by five G? There's a lot of hype about five short the moment and not just five jail. So things like WiFi six. Yeah, it would appear to me that this kind of technique would work equally well for five g Your wife. I short a WiFi six. So what are some of the use cases? You see these thieves service providers with Toko Edge clouds using this for? Yeah, So I think overall, first of all, I'd >> say enterprise use cases are going to become a pretty prominent part off five, even though a lot off the buzz and hype ends up being about consumers and how much bandwidth and data they could get in or whether five chicken passing preys or not. But in fact, things like on premise radio on whether that is private. Lt it's 40 or five t. These are the kinds of Uschi cases that were actually quite excited about because these could be deployed literally today. I mean, sometimes they're not regulated. You can go in with, like, existing architectures. You don't need to wait for standardization to break open a radio architecture. You could actually do it, Um, and >> so this sort off going in and >> providing connectivity on an enterprise network that is an enhanced state off where it is today. We've already started that journey, for example, with yellow cloud and branch networking. Now, if we can take that toe a radio based architecture for enterprise networking, So we think, ah, use case like that would be very prominent. And then based on edge architectures distributed networks now becoming the next generation Cdn is an example. That's another application that we think would be very prominent. And then I think, for consumers just sort of getting things like gaming applications off on edge network. Those are all the kinds of applications that would consume this sort off high skill, reliability and performance. >> Can you give a little sketch of the company pre acquisition, you know, is the product all g eight? How many customers you? Can you say what you have there? Sure >> it does us roughly three years old. The company itself so relatively young. We were around 33 people total. We had a product that is already deployed with chairman Telcos. So it is in production deployment with Chairman Telco Ondas in production trials with a couple of other tier one telcos. So we built a platform to scale to the largest networks in the world on If I, if I were to summarize it, be basically can observe, makes sense or in real time about every user in the network, what their experiences like actually apply. I modeled on top of that to optimize each user's expedience because one of the vision bee had was the network today is optimized for the average. But as all off our web expedience personalized netbook experiences, not personalized can be build a network Very your experiences personalized for you for the applications, your running on it. And this was kind of a foundation for that. >> I mean, we In fact, as we've been deploying our telco Cloud and carrier networks, we've also been counting roughly how many subscribers are being served up. Today we have over 800 million subscribers, and in fact, I was talking to someone and we were talking about that does. Being over 10% off the population of the world is now running on the lack of memory infrastructure. And then along comes Johanna and they can actually fine tune the data right down to a single subscriber. Okay, so now you can see the sort of two ends of the scale problem and how we can do this using a I. It's pretty powerful. Excellent. >> So So if we have any problems with our our service fighters, b tech support and I love to hear from both of you, you know what this acquisition position means for the future of the places and obviously VM wear global footprint. A lot of customers and resource is. But you know what I mean to your team in your product. >> I mean, definitely accelerating how quickly we can now start deploying. This and the rest of the world be as a small company, have very focused on a few key customers to prove the technology we have done that on. I think now it's the face to scale it on. Repeat it across a lot of other customers, but I think it also gives us a broader canvas to play that right. So we were focused on one aspect of the problem which is around, if you will, intelligence and subscriber experience. But I think with the cloud on but the orchestration products that are coming out of the ember, we can now start to imagine a full stock that you could build a network of full carrier network code off using using remote technology. So I think it's a broad, more exciting, actually, for us to be able to integrate not just the network data but also other parts of the stock itself. And >> it strikes me that this probably isn't just limited to telcos, either. The service providers and carriers are one aspect of this bit particularly five G and things like deployments into factory automation. Yes, I can see a lot of enterprise is starting to become much in some ways a little bit like a tell go. And they would definitely benefit from this >> kind of thing. Yeah, I mean, in fact, that's the basis of our internal even bringing our telco and EJ and I ot together and a common infrastructure pool. And so we're looking at that. That's the capability for deploying this type of technology across that. So you're exactly right, >> Checker want to give you the last word, you know, Telco space, you know? And then, obviously the broader cloud has been, you know, a large growth area. What, you want people taking away from the emerald 2019 when it comes to your team? >> Yeah, I think. To me, Calico's have a tremendous opportunity to not just be the plumbing and networking providers that they can in fact, be both the clowns of tomorrow as well as the application providers of tomorrow. And I think we have the technology and both organically as well as through acquisitions like Ohana. Take them there. So I'm just super excited about the journey. Because I think while most of the people are talking about five D as this wave, that is just beginning for us, it's just a perfect coming together on many of these architectures that is going to take telcos into a new world. So we're super excited about taking them. >> Shaker. Thank you so much for joining against auction. Congratulations and good luck on the next phase of you and your team's journey along the way. Thank you. Thank you for Justin. Warren comes to Minutemen, Stay with us. Still a bit more to go for VM World 2019 and, as always, thank you for watching the Cube.

>> live from the Mandalay Bay Convention Center in Las Vegas. It's the king covering via World 2016 brought to you by IBM Wear and its ecosystem sponsors. Now here's your host stool minimum, >> including I Welcome back to the Q bomb stew. Minuteman here with my co host for this segment, Mark Farley, and we'll get the emerald 2016 here in Las Vegas. It's been five years since we've been in Vegas, and a lot of changes in five years back Elsa do this morning was talking about five years from now. They expect that to be kind of a crossover between public Cloud becomes majority from our research. We think that flash, you know, capacities. You know, you really are outstripping, You know, traditional hard disk drives within five years from now. So the two guests I have for this program, Brian Vials, is the CEO of Day Tree. Um, it's been a year since we had you on when you came out of stealth on really excited cause your customer along. We love having customers on down from Alaska, you know, within sight view of of of Russia. Maybe on Did you know Ben Craig, who's the c i O of Northern Bank. Thank you so much for coming. All right, so we want to talk a lot to you, but real quick. Ryan, why do you give us kind of the update on the company? What's happened in the last year where you are with the product in customer deployments? >> Sure. Last year, when we talked, daydream was just coming out of stealth mode. So we were introducing the notion of what we're doing. Starting in kind of mid Q. One of this year, we started shipping and deploying. Thankfully, one of our first customers was Ben. And, uh, you know, our our model of, ah, sort of convergence is different from anything else that you'll see a v m world. I think hearing Ben tell about his experience in deployment philosophy. What changed for him is probably the best way to understand what we do. >> All right, so and great leading. Start with first. Can you tell us a little bit about north from bank? How many locations you have your role there. How long you've been there? Kind of a quick synopsis. >> Sure. Where we're growing. Bank one of three publicly traded publicly held companies in the state of Alaska. We recently acquired residential mortgage after acquiring the last Pacific Bank. And so we have locations all the way from Fairbanks, Alaska, where it gets down to negative 50 negative, 60 below Fahrenheit down to Bellevue, Washington. And to be perfectly candid, what's helped propel some of that growth has been our virtual infrastructure and our virtual desktop infrastructure, which is predicated on us being able to grow our storage, which kind of ties directly into what we've got going on with a tree and >> that that that's great. Can you talk to you know what we're using before what led you to day tree? Um, you know, going with the startup is you know, it's a little risky, right? I thought, Cee Io's you buy on risk >> Well, and as a very conservative bank that serves a commercial market, risk is not something that way by into a lot. But it's also what propels some of our best customers to grow with us. And in this case, way had a lot of faith in the people that joined the company. From an early start, I personally knew a lot of the team from sales from engineering from leadership on That got us interested. Once we kind of got the hook way learned about the technology and found out that it was really the I dare say we're unicorn of storage that we've been looking for. And the reason is because way came from a ray based systems and we have the same revolution that a lot of customers did. We started out with a nice, cosy, equal logic system. We evolved into a nimble solution the hybrid era, if you will, of a raise. And we found that as we grew, we ran into scalability problems. A soon as we started tackling beady eye, we found that we immediately needed to segregate our workloads. Obviously, because servers and production beauty, I have a completely different read right profile. As we started looking at some of the limitations as we grew our video structure, we had to consider upgrading all our processors, all of our solid state drives, all of the things that helped make that hybrid array support our VD infrastructure, and it's costly. And so we did that once and then we grew again because maybe I was so darn popular. within our organization. At that time, we kind of caught wind of what was going on with the atrium, and it totally turned the paradigm on top of its head for what we were looking for. >> How did it? Well, I just heard that up, sir. How did the date Reum solution impact the or what did you talk about? The reed, Right balance? What was it about the day trim solution that solved what was the reed right? Balance you there for the >> young when we ran out of capacity with our equal logic, we had to go out and buy a whole new member when he ran out of capacity with are nimble, had to go out and buy a whole new controller. When we run out of capacity with day tree, um, solution, we literally could go out and get commoditized solid state drives one more into our local storage and end up literally impacting our performance by a magnifier. That's huge. So the big difference between day trim and these >> are >> my words I'm probably gonna screw this up, Bryant, So feel free to jump in, and in my opinion day trip starts out with a really good storage area network appliance, and then they basically take away all of you. I interface to it and stick it out on the network for durable rights. Then they move all of the logic, all of the compression, all of the D duplication. Even the raid calculations on to software that I call a hyper driver that runs the hyper visor level on each host. So instead of being bound by the controller doing all the heavy lifting, you now have it being done by a few extra processors, a few extra big of memory out on their servers. That puts the data as close as humanly possible, which is what hyper converging. But it also has this very durable back end that ensures that your rights are protected. So instead of having to span my storage across all of my hosts, I still have all the best parts of a durable sand on all the best parts of high performance. By bringing that that data closer to where the host. So that's why Atrium enabled us to be able to grow our VD I infrastructure literally overnight. Whenever we ran out of performance, we just pop in another drive and go and the performances is insane. We just finished writing a 72 page white paper for VM, where we did our own benchmarking. Um, using my OMETER sprayers could be using our secondary data center Resource is because they were, frankly, somewhat stagnant, and we knew that we'd be able to get with most level test impossible. And we found that we were getting insane amounts of performance, insane amounts of compression. And by that I can quantify we're getting 132,000 I ops at a little bit over a gig a sec running with two 0.94 milliseconds of late and see that's huge. And one of the things that we always used to compare when it came to performance was I ops and throughput. Whenever we talk to any storage vendor, they're always comparing. But we never talked about lately because Leighton See was really network bound and their storage bender could do anything about that. But by bringing the the brain's closer to the hosts, it solves that problem. And so now our latent C that was like a 25 minutes seconds using a completely unused, nimble storage sand was 2.94 milliseconds. What that translated into was about re X performance increase. So when we went from equal logic to nimble, we saw a multiplier. There we went from nimble toed D atrium. We saw three Export Supplier, and that translated directly into me being able to send our night processors home earlier. Which means less FT. Larger maintenance window times, faster performance for all of our branches. So it went on for a little bit there. But that's what daydreams done for us, >> right? And just to just to amplify that part of the the approached atrium Staking is to assume that host memory of some kind or another flash for now is going to become so big and so cheap that reads will just never leave the host at some point. And we're trying to make that point today. So we've increased our host density, for example, since last year, flash to 16 terabytes per host. Raw within line di Dupin compression. That could be 50 a 100 terabytes. So we have customers doing fairly big data warehouse operations where the reeds never leave the host. It's all host Flash Leighton see and they can go from an eight hour job to, ah, one hour job. It's, you know, and in our model, we sell a system that includes a protected repositories where the rights go. That's on a 10 big network. You buy hosts that have flash that you provisions from your server vendor? Um, we don't charge extra for the software that we load on the host. That does all the heavy lifting. It does the raid compression d do cloning. What have you It does all the local cashing. So we encourage people to put as much flash and as many hosts as possible against that repositories, and we make it financially attractive to do that. >> So how is the storage provisioned? Is it a They're not ones. How? >> So It all shows up, and this is one of the other big parts that is awesome for us. It shows up his one gigantic NFS datastore. Now it doesn't actually use NFS. Itjust presents that way to be anywhere. But previously we had about 34 different volumes. And like everybody else on the planet who thin provisions, we had to leave a buffer zone because we'd have developers that would put a bm where snapshot on something patches. Then forget about it, Philip. The volume bring the volume off lying panic ensues. So you imagine that 30 to 40% of buffer space times each one of those different volumes. Now we have one gigantic volume and each VM has its performance and all of its protection managed individually at the bm level. And that's huge because no longer do you have to set protection performance of the volume level. You can set it right in the B m. Um, >> so you don't even see storage. >> You don't ever have to log into the appliance that all you >> do serve earless storage lists. Rather, this is what we're having. It's >> all through the place. >> And because because all the rights go off, host the rights, don't interrupt each other the host on interrupt together. So we actually going to a lot of links to make sure that happens. So there's an isolation host, a host. That means if you want a provisional particular host for a particular set of demands, you can you could have VD I next door to data warehouse and you know the level of intensity doesn't matter to each other. So it's very specifically enforceable by host configuration or by managing the VM itself. Justus, you would do with the M where >> it gets a lot more flexibility than we would typically get with a hyper converge solution that has a very static growth and performance requirements. >> So when you talk about hyper convergence, the you know, number one, number two and number three things that we usually talk about is, you know, simplicity. So you're a pretty technical guy. You obviously understand this. Well, can you speak to beyond the, you know, kind of ecological nimble and how you scale that house kind of the day's your experience. How's the ongoing, how much you after, you know, test and tweak and adjust things? And how much is it? Just work? >> Well, this is one of the reasons that we went with the atrium is well, you know, when it comes down to it with a hyper converge solution, you're spanning all of your storage across your host, right? We're trying to make use of those. Resource is, but we just recently had one of our server's down because it had a problem with his bios for a little over 10 days. Troubleshooting it. It just doesn't want to stay up. If we're in a full hyper converged infrastructure and that was part of the cluster, that means that our data would've had to been migrated off of that hostess. Well, which is kind of a big deal. I love the idea of having a rock solid, purpose built, highly available device that make sure that my rights are there for me, but allows me to have the elastic configuration that I need on my host to be able to grow them as I see fit. And also to be able to work directly with my vendors to get the pricing points that I need for each. My resource is so our Oracle Servers Exchange Server sequel servers. We could put in some envy Emmy drives. It'll screen like a scalded dog, and for all of our file print servers, I t monitoring servers. We can go with Cem Samsung 8 50 e b o. Drives pop him in a couple of empty days, and we're still able to crank out the number of I ops that we need to be able. Thio appreciate between those at a very low cost point, but with a maximum amount of protection on that data. So that was a big song. Points >> are using both envy. Emmy and Block. >> We actually going through a server? Refresh. Right now, it's all part of the white paper that way. Just felt we decided to go with Internal in Vienna drives to start with two two terabyte internal PC cards. And then we have 2.5 inch in Vienna ready on the front load. But we also plumbed it to be able to use solid state drive so that we have that flexibility in the future to be able to use those servers as we see fit. So again, very elastic architecture and allows us to be kind of a control of what performance is assigned to each individual host. >> So what APS beyond VD? I Do you expect to use this for? Are you already deploying it further? >> VD I is our biggest consumer of resource is our users have come to expect that instant access to all of their applications eventually way have the ability to move the entire data center onto the day trim and so One of the things that we're currently completing this year is the rollout of beady eye to the remaining 40% of our branches. 60% of them are already running through the eye. And then after that, we're probably gonna end up taking our core servers and migrating them off and kind of through attrition, using some of our older array based technology for testing death. All >> right, so I can't let you go without asking you a bit. Just you're in a relationship with GM Ware House Veum. We're meeting your needs. Is there anything from GM wear or the storage ecosystem around them that would kind of make your job easier? >> Yes. If they got rid of the the Sphere Web client, that would be great. I am not a fan of the V Sphere Web client at all, and I wish they'd bring back the C Sharp client like to get that on the record because I tried to every single chance I could get. No, the truth is the integration between the day tree, um and being where is it's super tight. It's something I don't have to think about. It makes it easy for me to be able to do my job at the end of the day. That's what we're looking for. So I think the biggest focus that a lot of the constituents that air the Anchorage being where user group leader of said group are looking for stability and product releases and trying to make sure that there's more attention given to que es on some of the recent updates that they have. Hyper visor Weber >> Brian, I'll give you the final word takeaways that you want people to know about your company, your customers coming out. >> Of'em World. We're thrilled to be here for the second year, thrilled to be here with Ben. It's a It's a great, you know, exciting period for us. As a vendor, we're just moving into sort of nationwide deployment. So check us out of here at the show. If you're not, check us out on the Web. There's a lot of exciting things happening in convergence in general and atriums leading the way in a couple of interesting ways. All >> right, Brian and Ben, thank you so much for joining us. You know, I don't think we've done a cube segment in Alaska yet. so maybe we'll have to talk to you off camera about that. Recommended. All right. We'll be back with lots more coverage here from the emerald 2016. Thanks for watching the Cube. >> You're good at this. >> Oh, you're good.