 The noise, it's theCUBE, covering VMworld 2015. Brought to you by VMware and its ecosystem sponsors. Now your hosts, John Furrier and Dave Vellante. Okay, welcome back everyone. We are live in San Francisco for VMworld 2015. This is Silicon Angles theCUBE, our flagship program where we go out to the events and extract the signal from noise. I'm John Furrier, my co-host Dave Vellante. Our next guest is Andy Warfield, CTO and co-founder of Coho Data. Welcome to theCUBE. Thanks guys. How you feeling? It's day one, you got to get settled in. I'm still finding my feet on VMworld, but it's been a pretty exciting day so far. So we've been covering Coho Data on Silicon Angle, Wikibon as well and the research side, you guys doing real well. Big time backers, you got Ignition and Andreessen Horowitz, you guys are growing. Here's the update. What's going on with Coho and technology? The company update, technology, and what are you guys doing here at VMworld? Wow, it's been a busy year. I think we're a platinum sponsor this year. We've got a massive booth down there close to the doors. We announced 30 million in our series C new funding earlier this year with HP and Intel as well as March participating in that. So some big industrial backers behind Andreessen and Ignition still participating. We announced our first all flash product. So earlier this year, we put out an all flash box that can seamlessly mix into the existing hybrid stuff that we sell. And then a couple of weeks ago, we announced that I believe we're the first storage company to allow you to extend the storage system by pushing containerized workloads down to run adjacent to the data in the storage. And so that'll be a big growth area for us technically over the rest of the year. What's the show like for you guys day one now? So yesterday opened it up. Great. Group traffic looked pretty big, strong. Super, super interesting. Lots of people around. The traffic's been steady and the people coming through the booth are really engaged. What are they asking you? What's the conversation like? I think it's, I mean, it depends on the size of the customer or the attendee that we're talking to. But there's a lot of similar themes through it. And it's sort of reflected the same kinds of conversations that I've had to the first half of the year coming up to the show. I think when we started the company, we spent a lot of time talking about really, really technical, gory stuff, right? I think a lot of the reason that we attracted initial investment was our work with really high performance flash with PCIe and our integration with OpenFlow and software defined networking as part of the storage layer. And I talked to some customers, we get really, really excited about it. Some customers were just totally glazed over when I started talking about, I get excited about SDN stuff. And so does Markeem, he comes up like, yeah. Exactly. The thing that we've seen this year is, the story for Coho with customers is one about just suffering through and embracing change in the enterprise, right? That I remember with the Martin stuff, I loved watching his interviews with you guys around SDN when it was really, really nascent and how disruptive it was. And we're still seeing that, right? SDN still, NSX has grown like crazy and they're into smaller accounts instead of just the really big stuff, but it's still taking a long time. And the reason that there's many storage companies out there as you see is storage is even harder in some senses, right? And so nobody really knows, I think, where a lot of this is going to pan out, where the show is going to pan out. And what areas is the hardest, you're saying? Because complexity, is it just more, I mean, obviously, storage or storage, but is it because of the complexity, because of the software side of it, customer deployments, what are some of those nuances? I think there's two sides. I think technically inside the engineering team with one device, right? Whether it's Fusion, Gear, Intel, PCIe Flash or some of the 3D cross-point stuff that Intel just came out with, I can saturate a NIC. And so I've just invalidated all the storage architectures you've had for 30 years, right? Raid behind that NIC is a waste of performance. Found the new bottleneck. Yeah, exactly, bottleneck shifted and it's a big deal. But the other half of it is owning that thing and having to replace it every three to five years and think about LUNs and volumes and file systems and all of that stuff, right? That's a thing that nobody wants to focus on. And I think that our story is probably similar to a lot of the upstarts in this space, that we are making it operationally easy and cheap to build complicated fast storage systems and that's a win. So you've got this cartel, we call it, and they're kind of happy with the LUNs but they talk the game that you're talking, but they haven't really delivered on those promises. But at the same time, they're able to sort of keep. Keep people at bay on stuff. Keep people at bay, freeze the market, okay. Is that starting to change? It is. People are saying, hey, wait a minute, I've heard this before, now I'm going to try, take a chance on somebody new. Talk about that dynamic. So I've seen this change amazingly even over the three or four years we've had the company. When we were starting, conversations around buying storage from a startup, right, it's storage. It's like, it's a big thing to take on. People have been much more embracing of the new entrance. Right, I mean, we've matured and gotten bigger over that same period of time, so there's both, but I think there's a lot more appetite and a lot more interest. The other one I think is really, really understated around companies like us that are doing a scale out product is one of the really, really big advantages of scale out is when I go in to compete with one of the larger incumbent storage vendors in a customer that has three, five, 10 petabytes of storage, I'm not trying to compete on an RFP to replace the whole thing, right? I'm not coming in at the end of their cycle. I'm having the conversation midway through and saying, I bet you've got some performance problems because invariably they do, right? Or some management problems, and we'll go in with a bit of run rate spent with one or two boxes. They'll try us out and then they'll come back a month later or a quarter later and start to move more and more workloads onto it. So they get to breathe life into the storage they own and they get to, you know, try something. They get to put their toe in the water. They don't need to come in and say, hey, rip and replace. Exactly. What's the thing there between latency and capacity? That seems to be a big thing right now. Yeah, yeah. So you want to be able to spend on both and you want to be able to spend on both as you need it, right? I think this has been a thing that some of the early flash vendors right in an effort to be able to hit a claim around all flash, you know, had to come on one side of. The first product we sold was hybrid, right? Mixed flash and disk. The AFA allows you to dial the performance end of that, you know, specifically in the capacity end separately. So when you say dial, that's something that the user dials or your system, I mean, you've injected analytics into your system. Yeah. Can you automate that for me? We absolutely do. So my brain can't handle all that. It's been an interesting thing for me. I mean, my background is with operating systems research. Was one of the guys that worked on the Zen hypervisor when we did that and Zen source before it or after it. We, early on with this, realizing that we were dealing with these really, really disparate types of storage, right? Spinning disks and high performance flash and working with the network. We wanted to borrow a lot of ideas from SDN, right? Have a system with a sort of central brain that dealt with placement and response to workloads. And so we actually brought on a small team of data science guys, right? Some stats guys and so on. We made them suffer through learning about storage and enterprise workloads for two years. And so we had a case with a customer a couple of months ago who called up, trying to understand some behavior of tearing in the system. And we came back to him two days later with a detailed profile of his 10 heaviest workloads and said, you make the decision on what you want to do with these, right? This is the stuff that's causing a lot of work on the system. And so being able to move storage into a space where you're talking to customers about workloads and he's not going like, do I have enough disks in this aggregate? So workloads are driving the conversation? Absolutely. So how does that change the game on the architecture side? And what can you share with the folks out there? Is the key driver? Because that's the top conversation we're hearing with customers. Apps are driving the conversation. Hybrid cloud is ultimately moving workloads around. So they want an easy under the hood, if you will, scalable solution. That seems to be like the big issue. How are you guys addressing that? And how do you talk to customers about that? You say, hey, you can buy as you grow. That's exactly what we say. So a box, is it software, is it on-prem? So I mean, as with several other, I think enterprise products today, we've moved to a model where we're packaging and reselling commodity hardware, right? The commodity hardware curve with things like open compute, the new switching, we work really closely with Arista on the open flow integration we do today, but things like Facebook's wedge and so on coming out. The massive change in terms of high performance flash, right? We want to let our customers not worry about all that stuff. Just take advantage of it. And so the team is dominantly software oriented. And we've released, I think, almost some form of hardware change every quarter in terms of making the product better. And they all interact into, you know, heterogeneous installs. So pretend I'm a customer for a moment. I got a big blank check. I need this. I got all those initiatives going on. I got a cloud native vision. I want to do hybrid cloud. I got a gun to my head. The business units want results, but I have all this existing stuff. What do I do? Can I mix and match? How does that work? How do you solve that problem? So with the legacy in particular? Yeah, with the legacy side, we see two conversations here, right? We see the conversation between the guy in the trenches that's building up the system that's got a big pile of incumbent vendor storage. And then we've got the conversation that he's having with his CIO, where the CIO is going, why don't you move everything to AWS, right? Why don't we go that way? And the conversation between the two of those ends up being actually remarkably similar in that the ownership and acquisition model for us is priced competitively with AWS. Scalo means we can price it operationally. You can buy and use what you need. Well, AWS has a lot of hidden costs as well. Absolutely. They can't forecast what your numbers will be. That's some level. That's totally true. That seems to be a game changer. I think that... Or a showstopper, if you will. On some level. On some levels, yeah. And so I think on both of those sides, being able to bring in-house a service that you own, that you directly manage and that you grow, hides a lot of those uncertainties from the public cloud side, but gives you something that's better in a lot of ways than the incumbents' storage offerings. I haven't changed that much over a long time. So can you talk a little bit more about the architecture? I mean, you get these interesting battles between the all flash guys and the HDD guys and even sometimes the hybrid guys saying, no, no, no, it's going to be a balanced world and all flash guys of course say, no, it's going all flash. The all flash data center, we've written about the all flash data center, maybe some tongue in cheek. And you guys, I think, just qualified a helium drive and you're going all kinds of deep direction as well as the all flash. So talk about the architecture and what's unique about it. So I mean, the architecture is specifically intended to mix and match, right? That is the core thing at its heart. The top tier of storage is the highest performance flash you can buy. It's NVMe flash, right? The individual devices will saturate a neck and the integration with SDN is to expose that. The end result is just push button scale, simple storage, right? We hide all of that complexity, but we allow you to price depending on whether your workload needs capacity or needs performance, right? We have the analytics to tell you what your workloads need. The most recent announcement that we've done around starting to allow containers to be pushed down into the system is really exciting to me because it's a much different take on containers than you're seeing from a lot of the hyperconverged and even VMware themselves, right? We're using containers in a much different way. We're using them from the perspective of data rather than as a repackaging of VM workloads that you would run on top of the hypervisor. So it's a data container. It is, well, no, it's a container for compute, but we're going to place it next to the data. So a good example of this, one of the first examples that we did was to work with media entertainment around a workload and this repeats all over the place with enterprise storage, you frequently see jobs where data comes into storage and immediately goes back out into a VM or a physical server to do something like transcoding, right? And then the results go back into the storage, right? That's something that doesn't necessarily need to live in an external VM, right? There's a ton of value in moving that compute close to the data, making the data active, if you will. Other examples include things like audit or security policies, right? Bringing able to bring big data jobs to run next to the data. So we're not really chasing the idea that you're going to host all of your VMs on it because in my experience with Zen, VMs are idle most of the time. They consume a lot of RAM, right? If we were to pack a ton of VMs onto our storage system, we wouldn't be able to expose the value of the expensive flash that we have. So they're heavier, right? Okay, so you got this container capability, it's lighter weight, you can move things faster. It's just simpler. Yep, absolutely. So what do you make of VMware's moves into storage? I mean, let's back up a little bit. With their VMware, you know, where are they headed? You know, imagine a world without EMC and how would VMware behave? Would they be more aggressive? Are you starting to see them sort of nibble away at the storage value chain as an ecosystem player? How do you perceive that? So I think, I mean, this is a question I think about all the time. Our architecture shares a lot of similarities and similar heritage to vSAN itself, right? We have very similar approaches to how we do a lot of the stuff that we do. Now, if I was running vSAN, I would be chasing way more than virtual machine storage, right? I'm building a product that's similar in senses to vSAN and I'm absolutely chasing general purpose storage workloads, big data, right, tons of other stuff. I have no idea to what degree vSAN's development is hamstrung by, you know, having other things around it. We could speculate, but you've got to us. But okay, so you're saying you're big a tam that you're playing in, right? Your general purpose. Absolutely. Not VMware specific, but of course we're at VMworld. So this is our, and you know, when we started the company, we wanted to focus on a use case that we understood and that we could build for in a short period of time. So we focused absolutely on the VM use case, right? Since then, VMware is our bread and butter, a little bit of open stack, but we're really, really aggressively broadening into the more traditional NFS, right? And big data workloads, right? And that's evolving because it's emerging, right? So you go for the beach head in VMworld, called Ecosystems, VMware's Ecosystem. And then kind of go with the market's development of the stack now. It seems to be developing. Are you happy with open stack right now? Do you see that developing in a way? It's a really, really interesting environment. There are some big, big users of it who build everything themselves and there are a bunch of small users who I think write code for it. I haven't found the sort of like mid-size open stack customer, it's not that they're not there, but we haven't seen them a lot. So you got to ask you, since you're a veteran in the industry and go way back to the Zen days, got to be proud of the results you guys have done with, I mean, look at AWS, you guys really enabled a lot of that. A lot of folks out there know that that core DNA was all open source and Zen was a big part of AWS success. But look at where it's at now. And I want you to think about this and share with the audience. Is there a hybrid cloud? I mean, does it really exist? I mean, I just retweeted Michael Cote, who's now at Pivotal, was an analyst from 800 days ago where he tweeted, oh, hybrid cloud, he's going 800 days ago. We're talking about hybrid cloud. Now, still it's a lucid. I mean, is it a product? Is it a category? Or is it just an outcome of how things are deployed? So I don't see a hybrid cloud product. What would that, I don't necessarily see a cloud product today. The environment. It's, yeah, I mean, it's tools on the shelf, right? Like as the public cloud has matured, you see storage products that take advantage of facets of the public cloud to do things like assisted replication or backup, right? As it evolves, you start to see on-premise, hosting things that are really geared toward Kolo rather than enterprise. I think that rather than saying that there's going to be a hybrid cloud product, I would see the notion of public and private going away, right? That hybrid cloud becomes the outcome of an integrated environment. Exactly. It's just like distributed computing. Building IT, it's, yeah. It's like, hey, I want to buy some distributed computing. Yeah. Where is that? There's two great examples of this that I've seen recently. One is if you look at the capabilities on the storage media side that are coming at us over the next three or four years, when we first worked with, you know, some of the early efforts around AWS was Zen, right? And we're working on the open source side and supporting that, turning into a big thing. There just weren't, like, yes, it was a challenging thing to build VMs at scale for an environment like that, but storage was actually a big problem. There weren't giant storage systems that you could take off the shelf and deploy for that. That was something that AWS really did a great job of solving, right? The whole ecosystem around S3, EBS, right? And so on. If you look at where the media is taking us today, within a few years, I'll be able to give you a two-U box that's in the neighborhood of a few million IOPS, right? I'll be able to give you a three or four-U box that's probably like four or six petabytes. Yeah, I mean, can moving compute to the data is the way people want to do it. I mean, self-driving cars could go there. It's interesting you bring up the Amazon thing, because I talked to Andy Jassy about this at length, and Dave and I sat down with him, and he talked about the early days of AWS where they were just going to build a storage service. Right. And the computers, a big debate around do we just do pure play siloed-like services? Right. And the answer was, they wanted to have an integrated set of building blocks. And so that was their decision. It turned out to be a good call. So you're saying today, same things happening now at a much larger scale? Well, and I'm saying, I think a bunch of the things, I mean, both in terms of performance and scale, and even functionality, right? With AWS starting to pioneer things like Lambda, protocol gateway, like things that you can't buy for the enterprise, right? I think that we're going to start to see a swing of applying sizing, if you will, that stuff to run in a colo, right? That- To emulate that. To emulate those types of things. That, you know, certainly with the Docker and container work that we're doing, we're supporting DevOps sort of template, where, you know, we hear from customers all the time that they want to be able to hire staff that developed S3, because they came through undergrad developing S3. And so that's why they wanted in the storage environment. And they want lightweight too. That's the key, another key thing. Application development, want a lightweight. Okay, so final question. I know we're getting the rap here. I got to interject though. One question, because we have the technologists here. A lot of that, of course, what AWS is building is homogeneous, right? And we heard VMware this morning talking about this hybrid cloud, but it's a heterogeneous world. We heard OpenStack, we heard, well, yeah, we'll connect to AWS and Azure. Pat Gelsinger has said that before. Can they achieve and emulate that capability in this heterogeneous environment? Or do you need that homogeneity? Ah, I think we'll, I mean, it's, is there a market for it, I think, at the end of the day, right? Well, sure. If people are running stuff across the various public cloud vendors, those APIs will exist to be, you know, brought into colo style appliances, right, I think. So it's a heterogeneous sort of coexistence. For the near term, I think, yeah. Exactly. Okay, final question for me then is end-to-end integrated systems, engineering systems, Oracle hardware, they're the iPhone for the enterprise. You're seeing VMware struggling with, hey, should we do puzzle pieces and put them together, Lego blocks? I'll, you know, service-oriented architecture kind of thing, or will there be a more end-to-end because what you guys are showing is, hey, the tool for the job doesn't matter whether it's an appliance or an environment, an engineered system for the job seems to be the trend. Yes. So that teases this end-to-end argument. Yeah. Are platforms end-to-end or vendors got to be end-to-end or both? I don't know. I love this question. I mean, I just wrote this blog thing last week kind of comparing what we're seeing in enterprise products to craft beer, right? That we used to have six beer vendors, right? And now there's thousands in North America. And I think that that's kind of what we're seeing, that there's a ton of opportunity to do a better job for a specific application. We've had a ton of really interesting early motion with the idea of moving applications closer to the data. It's different than a platform where you trust somebody else to do the VM work, right? It's a much bigger opportunity to partner and integrate with a data analytics company or whatever. I think that there's a ton of opportunity to work on that side and build packaged offerings on start. Andy, thanks so much for coming. I think you really appreciate it. Final, final question. What is your favorite craft beer? Oh, wow. I like a Vancouver one. I'm from the Canadian end of the West Coast. West Coast or East Coast? Best beer. West Coast for sure. For sure, yeah. Craft. Again, Silicon Valley, kind of aviation. We're talking about craft. That's the originator. Kind of where golf started, right? We are here now in Scotland. We are here in theCUBE live in San Francisco. We'll be right back with more of this short break. This is theCUBE. We'll be right back.