 Hi everybody, we're back. This is Dave Vellante. I'm with Wikibon.org and this is the Cube Silicon Angles Continuous Production. We come to the events like this. We extract the signal from the noise. We go out, we find the best guests that are at these events. We bring them to you, we package up that information to you, our audience. I'm at Dave Vellante, you can tweet me. My next guest really isn't a guest, he's a cube hardened, frequent co-host. The storage alchemist, Steve Keniston, is here. Steve, welcome back to the Cube, my friend. Thanks Dave. It's always a pleasure to come here and sit here and talk to you guys. I love what you guys do, the whole John Furrier extracting the signal from the noise. Pretty fantastic. You know, we're at this live event edge. A lot of new information coming out, a lot of topics that are being talked about. I'm interested to see what you guys think, what you're hearing, what you like, what's going on. So I think, I will tell you, there's so much going on in the storage business. I've been saying a couple of times this week, you've got sort of a dupe on one end of the vice, and you've got AWS on the other, and it's like got storage and compute and networking in its sites, and it's squeezing things to the max. I think those types of disruptions, people will often think, okay, the old line, the old guard is screwed. I think, well, that may be true. Much of the old guard is going to respond, and I think you and I have talked about this a lot. The IT business is this oligopoly, and you've got five big, six big players that control the chess board. They have huge piles of cash, they're really smart. They know how to make acquisitions, and they do so in a way that maintains their dominance in the industries, and I see that combined with some of the visionary opportunities that I've heard this week, particularly from Ambuj Goyal, as really exciting for IBM. It's not been this excited about IBM's storage business, I think, ever, to be honest with you. Yeah, I think it's a really exciting time, and you make a really good point, and I like your analogy. You've got the hoop on one side, and you've got all the services and storage, and you've got the compute on one side and acting as a vice, but the thing that's tying it all together, and we're hearing a lot of this, and I'm sure you're hearing a lot of this, it's the new industry buzz term, software defined, and software is going to help play a part and tie these things together to make each one of those pieces more valuable so that we can then become or generate competitive advantage. Yeah, so we're going to get into some of that here in the next 15 minutes or so. I think that the interesting thing around storage as a platform, I think, is worth spending some time on. I've said a couple of times this week, Amazon turned the data center into an API, and Amhush Goyal wants to turn storage into an API, a storage platform into an API. We're going to bring in Randy Arsenal into the discussion. Randy is another long-time CUBE guest. Randy, thanks for stopping by. We just saw you. We pulled you out of the crowd and said, hey, come on in, and we're very much pleased that you're going to participate in this conversation. Thanks, Dave. Appreciate it. Good to be here. Yeah, so just kind of a freewheeling discussion here. Let's sort of unpack. Let's start with SDS, software-defined storage. It comes out of this whole meme that I think VMware really started the software-defined data center. They made that NYSERA acquisition, a billion-dollar acquisition of a company that really didn't have a product in the marketplace, and that sort of got people's attention, kind of ticked Cisco off a little bit, and that became quite interesting. And then people realized that software-defined is actually a vision that we could actually realize today. So why start with Randy? Why software-defined and why now? Well, I think there's been, it's almost kind of a convergence of a number of things that are happening, not just things like acquisitions in the marketplace by vendors like VMware and others, IBM obviously as well. But I think if you look at the evolution of the gear itself, you've got much more software content. We've gone from a lot of proprietary hardware architectures to more open platforms, and the innovation is taking place in the software stack because it's faster, time to market is better, you can roll releases more easily, you can add functionality in a much more modular, kind of proven and repeatable way. So by adding functionality as services in the form of software, it just makes the whole platform more adaptable, more agile, and easier to apply to a wide range of use cases. So you're not locked in architecturally to a given set of characteristics or workload behavioral characteristics. So the whole movement toward software-defined X, whether it's storage, infrastructure, networking, it's kind of a logical evolution in my view. I think it was going to happen regardless. And as we see more and more processor horsepower available and more and more thread density and dye density, software becomes where the innovation takes place and where the advanced capabilities exist. So Steve, we talk about services. We try to throw that term around a lot. But so you've been with a number of companies who have provided appliances that have services in them. I think of store-wise with compression, that was a service, Avamar had a dedupe appliance, that was a service. These are the types of services that we're talking about. In addition to copy, replication, is that right? Yeah, exactly. I think a lot of customers want to be able to abstract the particular hardware for whatever the reason is. It might not be what their infrastructure really can handle or how they want to handle it. To some degree, selling that capability. And as you know, both with store-wise and with Avamar as a startup company, you've got to be able to provide a customer a full solution before they kind of trust that the software itself is going to take off and be flexible enough to run in their environment. However, when you get into the big companies after EMC, acquired Avamar, after IBM acquired, store-wise, you start to see us pull those services or those, and I like to call them storage services out of those devices and make it more open. So if you look at what we've done with real-time compression, taking it, putting it inside SVC, putting that service inside of the V7000 or the store-wise platform, for example, and we've heard here at the show a number of customers, including Rico, who got on stage saying they've tested it, it works, it works in their environment, they're seeing tremendous value, and it's really that value which is that next layer of what these services are providing that customers really need or they want. So these services have been locked in middleware, close to the controller and really proprietary and having metadata that's sort of hidden away. What, so you gave the example when store-wise got acquired, what happened? I mean, did you have to, was there an API that you guys were able to access or did you not do that because you're internal to IBM, and what I'm trying to get to is, is the marketplace model going to kind of mirror what happened internally at IBM to actually exploit those services and port them to other platforms? Can you guys talk about that a little bit? That's something that I don't know anything about. So I think what's interesting, the evolution of the technology is actually an interesting story. So when store-wise first came out in the marketplace prior to even me joining the company, you saw proprietary software code almost like firmware embedded in an appliance. Now at this time it was for NAS only and you bought this appliance and it was this embedded capability. So there was no way to really extract that. The only way you could obtain that service, and again this goes back five or six years when the company first started, is to purchase the appliance. Then what happened is we joined, we said we need the flexibility. We did this when we got to Avlar, Ed Walsh and myself. It was, how do you then put that, that software on top of some code a la Linux, right? And then embed that Linux platform inside of now we don't have to be so proprietary about the hardware. How do we make the hardware more open? So an IBM server was the next step for store-wise, the software. So an operating system with the code on top of a standard system could have been Dell, could have been we chose IBM. And then once you're inside of IBM now that that software is much more portable port that technology onto different platforms that now run this openness. So, so really I wonder if you have an angle on this. So we've been hearing a lot about Linux and we all remember of course the LAMP stack and what impact that had on the industry. Are we seeing a similar type of critical mass now in storage with regard to open source? I think we're kind of at an inflection point in a sense where it is starting to get to that point. And I think Anbuj made a great point in his presentation yesterday as keynote that no single vendor is in a position to be able to provide the necessary services and the richness of functionality and capability that the industry demands. We've got these very dynamic workloads. Cloud is driving a lot of change and a lot of volatility in the workload mix and the characteristics of these workloads. So there's not going to be any one single vendor that's going to be able to provide that robust and rich set of services that'll be necessary in this kind of next generation. So this collaborative industry-led approach which has proven very successful in the past with the examples that you mentioned is a sensible way for this to evolve. We're at a point now where the vendors, the software vendors, the solution providers, the vertical industry providers are starting to adopt a more open standard approach. Obviously the vendors and the infrastructure players are now starting to get on board with it. So I do think we're going to see a lot of acceleration towards that more open, more industry-led portfolio and environment. And frankly, I think it's going to cause challenges for proprietary vendors who are trying to force their own kind of reference architectures out into this ecosystem which is emerging very rapidly right now. You know, to Randy's point, it's what the customers want. That openness, that ability to pick best of breed, that's what they want. Okay, but historically you guys are a proprietary vendor trying to push your own code. So, of course, Booz often references Lou Gershner's line that you were covering alcoholic. Right. But so let's talk about that a little bit. So you've got staying on the theme of software defined. You've got this ability to abstract the hardware, virtualize the backend hardware through SVC. And then you've got this other piece on top which is going to be either be OpenStack or it might be VMware, could be Hyper-V, could be CloudStack and whatever platform is on top and you've got these north-south bound APIs there. Where do the storage services come from? Is that they come from SVC? Do they come from the proprietary box that sits at the backend? Is it new services developed? How does that all shake up? Well, sorry. Just quickly, I think you're seeing and we've been talking about it in the context of this architecture. There's sort of the data plane and the control plane. So the data plane is really what we've been doing for over a decade with SVC. So we pretty well understand how to abstract the storage devices and deliver those fundamental kind of atomic storage services. Then if you think about layering OpenStack or some of these other standards that are emerging on top at the control plane layer, it then becomes just a connectivity problem. Like how do you make sure you build the connections between the backplane and the front end? And that's where a lot of the engineering work is going right now. You heard Ambuja mention that we're contributing to OpenStack. We have got 250 developers that are actively contributing. We're a platinum sponsor. So we're building the bridges between the primary services and the front end. So that data, but that data plane, so your answer essentially pointed to SVC as most of those services will reside. But your other products have services too. There are other products in the industry that have. So we had one of your clients on Karim Abdullah from Sprint was saying, we use both. We use sometimes the SVC services and sometimes we use the services that are on the array. Do you see that change? In fact, when I had Brian Gallagher on at EMC World, we were talking about this and he said my strategy is to move faster, do a better job of delivering services than the open source we can now. Long term, that's tough strategy. That's not terribly sustainable. Short term, maybe it's a pretty good approach. How do you see that all shaking out? I mean, personally, I think long term open source wins, but today I can solve problems with the more proprietary approach. What's your take on that? So you hit the nail on the head, right? So yes, right, is the answer to your question, where do all these services come from? I think the store-wise story is a piece of that. We will acquire technology, we will continue to build technology, and where it makes sense, and a big part of that is when I say where it makes sense, we will provide those services heterogeneously across everything that we can, right? And it'll be driven by the customer demand. So for example, and we'll come back to compression because it's an easy example, right? Customers do not, you know, they want us, they want to, Randy puts us in a very good light, right? They want to decelerate the growth of storage in their environment, and they don't want to decelerate the growth of data because it's data that adds the value. They want to decelerate the growth of that footprint of the spinning disk. So how do you do that? Well, compression is the real-time compression technology is the best in the industry, right? It's transparent, it's fast, there's no impact of performance. Customers are standing up and saying, I want that. So you're not going to get that service in a particular array or a set of different arrays or want to manage it differently across different arrays. If you can provide a layer that lets you do that heterogeneously. You're saying it's not fungible. If you get it in just the array, you can only use it on that array. You can't, you know, spread it across, spread it across multiple sources of management then, right? So it becomes difficult. And customers again don't want difficult. And Randy, you were talking about the control plane. The control plane, is the control plane, for example, OpenStack or is that something that you guys provide or is it a combination? Well, it could be, I think it'll be a combination, right? It'll evolve along a couple of paths, right? So we see OpenStack really gaining an early leader, taking out a leadership position early on. There will be others that will emerge. And to your point you made earlier, it's very valid. Vendors will continue to provide their own proprietary solutions because they can get it into the market, they can cultivate their customer base, right? So they can go deliver these solutions today and do so in a way that delivers immediate value for their customers who might not want to undertake the engineering effort required to start looking at something like OpenStack. So we don't have a ton of time left, but I want to talk about data protection as a service. We've looked at that market, we all know EMC dominates the appliance market and it's like now we have this huge disruption going on. I'm sure they see it, you guys see it. How do you turn that into an opportunity? What do you see, Steve? Something that you know a lot about is the data protection business. What do you see as the opportunity there? Well, and again, Dave, you and I have talked about data protection for a long, long time. We talk about it being broken, we talk about what customers want. We've often talked about the time machine for the enterprise, but the technologies and capabilities, the combination of the ability to do continuous data protection, to be able to roll that into snapshots, to be able to manage that more effectively, to be able to leverage embedded storage capabilities like compression and deduplication are now making that set of capabilities real. When CDP came out, you had seven times the amount of storage of disk capacity you did. Who wanted seven times more disk just to get CDP, right? And add all that capacity to the data protection infrastructure. Nobody wanted that, but if you needed that type of recovery, you had to make the sacrifice. Today, those things aren't happening. So you're looking at new capabilities and new markets where this notion of being able to roll back inside the array, I look at it as as you build out your new LUN, for example, in your storage array, you're defining a storage service level and that service level is directly tied to a protection model. So no longer are you installing apps, moving data, it's funny, I asked a question in my customer or in a presentation I did yesterday. What's the most expensive thing to do in the data center, right? And the reaction I was looking for was move data, but the customer said data protection. So I took a step back and I said, well, wait a minute, that is moving data, right? Lots of it. So in a sense, he answered the question and customers don't want to do that. It's very, very expensive, right? So how do we start to alleviate that and there are new capabilities coming along to do that? In fact, we're demoing some of them here at Edge. All right, gentlemen, we're out of time, but Randy, we're going to give you the last tip. We need to talk about flash, but Randy, we'll give you the last word. We're here a year from now. What's changed? What are we talking about? What's your prognostication? Put on the- Well, I think flash is probably not a bad headline because I think we're clearly going to see accelerated adoption of flash. I think one of the most significant things we've done so far this year in terms of making a sort of disruptive statement in the industry is the economic characteristics of flash that were mentioned in the April 11th announcement. I think we're going to start to see that really come to fruition. We're already seeing rapid uptake in the marketplace within our customer base and competitive accounts as well because it just makes economic sense. We know it's fast. We know that it delivers great performance characteristics. It's hardened, it's reliable, it's durable, it's resilient, but now with this new economic model on top of it, it's becoming a much more attractive platform for the delivery of service. And in fact, it's opening up opportunities for software developers to architect new solutions to take advantage of it and think differently about how they're delivering their services. So it's really opening up a lot of interesting opportunities. You guys are always fun guests. I really wish we had more time. Thanks, Steve and Randy for stopping in. Thanks, Dave. All right, keep it right there. Wilfredo Sinalongo is here. We're going to talk system X. We'll be right back. This is theCUBE from Edge.