(music) >> Hello and welcome to theCUBE, here in our studio in Palo Alto, California. This is a CUBEConversation; I'm John Furrier, the co-founder and co-CEO of SiliconANGLE Media, and co-host of theCUBE. Our next guest here is Surya Varanasi, who's the co-founder and CTO of Vexata, a hot startup here in Silicon Valley, also exhibiting this week at Oracle Open World in San Francisco. It's our 8th year of coverage at Oracle Open World, we will not be there on the ground with theCUBE; not a lot of room as they're doing a lot of reconstruction up there, among other events happening. Great, great conversations happening around the world of cloud, and certainly Big Data, now called 'data' generally because it's so hot. Sorry, welcome to the CUBEConversation. >> Thank you. >> So, first of all you guys are a hot new startup, really coming out of stealth, but not really stealth I mean stealth technically, not with general availability. You've been in business for a few years, building up great comprehensive storage-slash-data solution, I call it, with this "data fabric" concept. Congratulations. >> Thank you. >> Well-funded team, super technical. Tell us about the company, tell us about the launch, you guys are out public at Oracle Open World this week. What is Vexata? >> Vexata, we started in 2014, as you mentioned, a few years in development. We've been in trials for over a year and a half, shipping actually for a good eight or nine months. And what we're about is, we really wanted to design against three basic pillars. The first one being, there's digital businesses, they're all under pressure. "How do we survive, and how do we handle the transactions that are coming in?" And we wanted to build the highest performance storage system that we could build that really accelerates your apps, makes them super fast. The second thing we want to do is, the demanding enterprises, these are the ones that have the requirements, we wanted to be super enterprise-resilient. How do we deploy seamlessly? That was the third pillar we stood on, meaning no changes to your application or your, how do we plugin and just simply work? So we simply work, we're enterprise-resilient, and we have the highest-performance system that accelerates your apps with no changes. We built around Flash, and Intel's latest 3D Optane, and that's a big deal. >> Well you guys are well-funded, looking at the management team of the company, it's a start-up that began a couple years now, but you're now out in the wild, now launching. Just a couple of stats here, over 50 million dollars in funding, well-funded, great with a lot of work on the front end with the product, but the venture capitalists are interesting. Lightspeed has been very, very successful in the enterprise, just look at the list of successful day value, even if they have Snapchat too, so they know a little bit about data. Mayfield, Intel Capital, and Redline, and International One. Really really good pedigree there, and they know storage too, they see you telling us, they know what it looks like, they understand converge investors they now get the data play. I have to ask you, in the market everyone's kind of scratching their heads right now because real time data is super important. What problem are you guys solving because certainly the performance >> Yeah >> Is looking good. What's the problem you solve for customers? >> So, the specific problem is when you have digital businesses, what happens is that you don't have a little bit of data that's hot and the rest that's cold, everything is hot, so how do you serve all your data in real time? That's what we're about, and that's what we've built a transformative solution for. >> Well the thing that's coming out, some of the feedback we've been getting and seeing online is, besides the new logo, looks great by the way >> Thank you >> Is you guys are winning on the speeds and feeds. Now the market's going beyond speeds and feeds, which we'll get to in a second, so one: talk about the performance goals, you guys are saying exponential performance, but you're saying you're 10x the performance of anything else. But two, the challenge with data is these silos, right, and you're seeing a confluence of injection of open-source coding, real-time performance data in the application level as app developers start to come on board with open source. At the same time, data traditionally has not been open and free, democratized, if you will, that it's stuck in silos. So it's been a big challenge for architects and CXOs to say, "How do we deploy a solution that gets us to value quickly, not do these science projects." So talk about the performance and then the market model around "How do you free the data?". >> Yeah, so I think for us the simplest of value props is when you plug into your existing infrastructure, we show up to any OS just like a disc. We show up very simply like a disc, so any application that runs with Vexata powering the disc, the virtual disc, if you will, runs enormously fast. That's the very simple value prop, we've done something very basic. >> So on the integration site, like deploying it's easy. >> Not only is it easy, there's no change to the OS. So you talk about democratization, what are you looking for? Can I simply plug and play, will this just work? So that's the biggest thing of all, it just works. The second piece, and the most important thing is, it's not just out here our numbers that really work well, when you plug in Oracle and run OLTP or OLAP, you see this dramatic performance that if you didn't know better, you'd think this was an engineered system from Oracle, you know it's really just amazing performance. We maximize the utilization of your server, so any app that just plugs into an OS and looks at it as a disc will run great. >> Well when you say that, not to trivialize this, because I know it's probably complicated, I'm going to dig into the tech in a second, but when I plug in a thumb drive or an external hard drive into my Mac, it's just "Boom there it is!" >> Yeah. >> Similar, is that the kind of concept you guys are thinking? >> Pretty much, that's what we, really if you build a very complicated product that's complicated to use, nobody'd use it. So we want it really simple to consume. Complicated to build maybe, but really simple to consume. >> Alright so I'm going to play the naysayer, I don't believe you guys, it's smoke and mirrors in there, Cause no one can do that, you're going to give me 10x performance? Okay, that's marketing, I'm skeptical, but I have a problem. I have I/O bottlenecks, at the end of the day I have all these bottlenecks, how do you do it? >> You know, I think a few core principles, the first of them being we use solid state media. How we read and write to that solid state media is actually under patent, it's very specific to keep the performance very high all the time. The second piece of course is our system itself is designed to avoid, to separate control and data paths, so we keep them isolated, and we've invested a lot in our software to keep it and use the space and so on, a lot of jargon for we keep our latencies extremely low on the system, so your applications don't have to worry about anything and change anything. >> So are you lower in the stack in terms of, well a stack isn't perfectly speaking but, I start thinking about free data moving around, which by the way, people want, they want their data available at any given time, at any moment, cause you don't know what's going on in real time. All the data has got to be ready. But then it brings up the governance thing. Are you below the governance or is that a separate challenge on top, how do you deal with that dynamic? >> I'd say we're in the governance of it, so you know for example we provide the full standard based encryption so should anybody say, "Hey are you secure?" Yes, absolutely we are. It's a big deal, it's data, it's your active data, and so we protect it as well. >> One of the things that's coming out of Oracle Open World we're seeing obviously is they're comparing themselves to Amazon. And I was commenting last night on Twitter, I've been covering Oracle since 1994, watching and comparing them against SAP back then, the ERP days, and all the software mini computer days. But now they're comparing themselves not to SAP or IBM anymore, it's Amazon. What does that tell you, because that's also translating into the customer conversations because cloud has become mainstage, Oracle says "We have the cloud, it's Oracle on Oracle." They're not really winning the Cloud Native battles, they kind of own IT, Oracle does, so there's really no debate that IT, information technology CIOs know all about Oracle, but people who are doing Cloud-Native or DevOps might not be interested in Oracle, so how do you balance those two markets that, Oracle's trying to be more Cloud-Native and we're still evaluating their progress there, but you guys, are you impacted by those trends at all? >> You know, as you mentioned, everybody talks about the cloud, a lot of apps do go to the Native Cloud, if you will, the data that's very critical to your business, be it your intensive transaction processing, your OLAP, your machine learning, those seem to remain on premise. That's what our experience has been, and that's where we want to play first. Now, Oracle for Oracle Cloud, Oracle Cloud for Oracle may be a great thing, but-- >> Oracle on Oracle runs well, but I mean they're still playing catch up to Amazon, clearly number one. Okay let's get, you bring up the on-prem thing, this is important, business model. So you guys are out there, share the business model for you guys. What's on premise, is it hardware, software, both? Is there license, how do people engage with you, what's your business? >> So today we sell an appliance, that's the product we have today, and so we sell the appliance all included, software and hardware, and we offer the services to plug it in, and show you the transformative results on your applications. We don't stop at "Hey we plugged it in and you got your hero numbers," we show you. >> So I'm, I just want to buy it, how do I engage? I buy a license? A box? >> You buy the system. >> System, so it's hardware. >> Yeah >> And all the software and the intellectual property that you have >> All in. >> Is inside the box. And how do I, just connect to the network? >> Pretty much. >> Like, all interfaces? >> Pretty much; so today we have fiber channel, and we have NVMe over Fabric, so both ethernet and fabric channel, this is typically where you're on your highest performance of your data. So for us, very simple, it's very seamless to plug it in, and it'll be recognized in your servers, and off you go. >> Okay, so I'm an architect at a large enterprise, take me through the conversation you'd have with those geeks because they're going to want to (Surya laughs) have the conversation be, I want it, I need dashboarding, we're going to be moving high value applications so I need analytics, I want to kill the memory bottlenecks, but I also want the future, I don't want to foreclose anything so, you know, you guys are a startup so you've got my attention, I like what your story is. How do we move forward in the future, how do you talk through the, we've got your back covered, you've got the head room available, how does an IT or tech guy say, "You guys are solid."? >> Yeah. So here's how I start the conversation: I typically start the conversation by telling them, "Hey, you've got the highest performance servers, the latest servers, the Broadwells, the Prolines, what have you. The fastest networks are the 100 GigE, 50 GigE, you know whatever your ethernet network looks like, and then typically you have a SAN and it's really fast- 1630 to a gig. And you run your application, let's pretend it's Oracle RAC, you run that application. And when you run it, what you notice in your servers is eventually you see I/O wait times that slow down your application, and your servers, your really fast servers are under utilized because they're just not moving. >> Because you have bottlenecks. >> That's right. Well we say, it's very simple, if you plug us into your network and run your application on us, we will eliminate your I/O bottlenecks on your server, so your server is maximally utilized. So with no changes for you, you get 10x, and that's how easy we want to make it. That's really our value. >> So you guys are coming in and basically saying 10x performance right out of the gate. >> Yeah. >> Okay so what are some of the challenges on the dynamics, because you got my attention again, now I say, "How do I know I need you? Is there certain things, smoke before the thing blows up?" What are some of the tell signs for the customers to call you guys, cause they just started hearing about you guys as you start marketing. Why should customers work with you, what's the indicators on their side where they go, "I got to call them." >> The classic indicator is, for us, for one is, you're running an Oracle RAC. You're running an Oracle RAC for resiliency and for performance and you need both, right? The moment we see that we say, okay, we have a clean in. The second tell-tale sign, is when you have in-memory databases running. When you're in-memory, what you're trying to do is not write to your storage because that's your bottleneck, so you keep throwing memory at it, it's really expensive. And we know that's a classic sign. >> Okay, talk about Oracle Open World, you're going to be here this week, up in the city. What are you guys showing, what's the pitch, obviously you've got the new logo. >> Oh yeah. >> @VexataCorp is the Twitter handle so people can watch and can check out your updates on Twitter, but what's the value proposition, what are people in the booth talking about, what's the demos, what's the thing? >> You know, it's Oracle Open World, so we're going to do a whole lot of Oracle demos, so we have a RAC demo set up, and we show, with a four node, dual socket server, our system seamlessly plug in, and you get, the last I looked, it was 4,000,000+ OLTP transactions. It's phenomenal for a four socket, dual socket server. We're going to show our optane base array, the first of its kind in the industry, and show the same kind of results we have with optane. So it's all about Oracle and accelerating those apps. >> Alright so for Oracle customers out there, you know who you are, they're always evaluating stuff but it's always hard to kind of get out on the branch and be exposed if you try to go off Oracle, so people might be a little bit nervous. What's your conversation to the Oracle customers that's saying there's no risk in looking at Vexata. They're like, hey why not just buy a lot more Exadata, or the ZDLRA stuff, or other things that they have. >> And all those are entirely possible, I think that's the easiest way to get comfort. It's those trials, even in your research and development, and get used to us, because you'll be shocked at the performance you get. And eventually yeah, you can go to Exadata, but we're just so much more cost-effective than any solution out there. Try us, get comfortable with us, and then deploy when you're ready. >> And what's the price point? What's the price, or is it different by deployment? >> You know, honestly, it does differ by deployment, but really we use standard NVMe flash, and that's driven by the HyperScale guys, so we ride the curve of flash, we don't make our own. >> Yeah, you're not a sales guy, you're a CTO, co-founder, >> Thank you. >> So I don't want to put you on the spot there. Affordability relative to Oracle, let's talk about the customer conversation, so I don't want to put you on the spot on the pricing, we'll hit the CEO and some of your other guys on that. So, I'm a customer, I'll roleplay. Hey, I love this opportunity, but what's wrong with Oracle storage, why not just go with those guys? >> You can use us, not just for Oracle, but all your application workloads that are demanding, like your machine learning, like your SQL server, if you're running SAS analytics, so you have a general purpose platform, you're not silo'd. That's the biggest deal with us. >> So you give them scope outside of Oracle. >> That's right. Any app, really, I mean, just the simplest of all. >> Alright so I got to ask you the secret sauce question. You've got some patents, so your friend says, "Hey, what's going on, you guys are awesome, how did you get the 10x?" What's the bottom line, how did you guys do it? How do you get all that performance? >> I think the really, the investment in the software, to reduce the storage stack latencies, to the absolute minimum, that's what really gives us the biggest bang for the buck. >> So a lot of low-level engineering. >> Pretty much. >> Alright, so benefits to customers? What are the benefits, how do you guys see the benefits unfolding, take us through some of the anecdotal data you've seen in the trials you've done with customers, what are some of the benefits they've told you they've seen? >> You know, the simplest of them all, it's a very simple one, when we do a PoC with a customer, the customer usually says, "Hey this PoC's going to take two months." And afterward we find out it's two months because it takes three weeks to tune the system, and then the remainder of the weeks to do the PoC. For us, those first three weeks collapsed to one day. There's no tuning, you just plug it in and you all of a sudden get the performance and your PoC has just shrunken massively. That's really our value. Don't try hard, just right to your data, embrace it and it'll run for you. >> So you guys are a potential bridge to the future with the data. >> Yeah. >> You have this thing called Active Data Fabric, is that it? >> Yes. >> What is that about? >> It's really about how you actually scale your data over a very large amount. Today, yes we have an appliance, and it scales on size, ours scales to 150 terabytes and so on, but as data keeps growing and everything becomes hot, you really need to get into the many hundreds of terabytes, petabytes of active data. So how do we actually design that using external, open hardware, that's really what the principle is about. So this is the first realization and then we continue going with other implementation. >> Surya, great to have you on theCUBE, you guys have done a great job, so I got to ask the bigger question outside of Vexata, you know, data's been a challenge, and as an industry participant and a technologist, what's been the big thing, if you could summarize it down from your perspective, data obviously needs to be free, because applications never know when a piece of data will be needed in context to other things. You see things like metadata, active data's clearly the benefit there, but everyone's got these data lakes out there. We just came back from our Big Data NYC event, and the whole Hadoop thing has been very batch. >> Yeah. >> Store everything in a data lake, but you never know, at any given time, if a piece of data is going to be valuable, until you put it to work. So you really can't put a valuation on data. What has been an inhibitor, the bottlenecks. Has it been the silos, has it been data architecture, has it been the software, or now that the cloud's got compute power, all of the above, what's your thoughts? >> I think you netted out, really data, you look at it as hot data or cold data, and you decide data lake or active data, and I hold it in memory. The biggest problem I see is how do you call something hot or cold, it's hard to tell, and I think the biggest challenge for us is how do you make it all at least warm, so you can get to it when you need to. And that's the hardest challenge for the industry, I think. >> Yeah and I think that people look at self-driving cars to bring up that, because Larry Ellison said onstage, "Autonomous database", which I kind of roll my eyes cause Larry's so good at taking trends and making it look like Oracle has it. Autonomous cars being self-driving, it's the concept. >> Yeah. >> The data's really critical, cause if car's going to have telemetry data, real time is real time, it's not milliseconds, it's nanoseconds. You can't say one week, ten days, and a lot of time people say realtime queries can come back, but the data's a week old, so there's huge issues in what real time means. >> Yeah, and the second issue, you bring up self-driving cars, so the way self-driving cars, the test drives happen today, you plug in a lot of drives into the car, send it out for two weeks, and when it comes back to base you have 200 terabytes in the car that you want to learn with. How long does it take to transfer 200 terabytes? In a regular system? A few days? >> Yeah. >> So until that data's off, this car doesn't move. With us, it takes a few hours, so you can get your car back on the road. So we actually, we not only do great on the transactions, we do great with this, your basic data mobility problems, and we fix it. >> Yeah, you guys are fixing the data mobility problems. Okay, great connotation, one last final point I want to get your thoughts and color on, the internet of things. Cause now you're seeing industrial really being the low hanging fruit right now on IOT, and IOT certainly is hot, it will always be hot, but it also increases the surface area for cyber attacks. So people are kind of taking baby steps there, first one is industrial: plant equipment, could be manufacturing, it could be edge of the network sensor, something along those lines, hacking into the IP network. That's certainly going to create the need for active data. >> Absolutely. >> Your thoughts on that? >> Very much so. You know, IOT is really the classic future growth model, look at the amount of data you're trying to ingest and process. Everything is active, and you have to act on it in real time. >> Does IOT help you guys? >> You know, it does, it quite doesn't show up as IOT, it shows up as machine learning, you get another signature off it, you get all this data, you're trying to learn and figure out anomalies, and you need to process your data and that's us. >> You know I always said that a good business model is reducing steps it takes to do something, making it easy to use, and being high performance. You guys seem to do all three, congratulations. >> We do, thank you. >> Surya Varanasi, CTO and co-founder of Vexata, hot new startup, check them out, @VexataCorp is the Twitter handle, check out their updates, they're at Oracle Open World this week. I'm John Furrier, you're watching CUBEConversation here live in our Palo Alto Studios, thanks for watching. (music)