 It's theCUBE, covering VMworld 2015. Brought to you by VMware and its ecosystem sponsors. And now your host, Dave Vellante. Since the dawn of the cloud. Hi, we're back. This is theCUBE and theCUBE goes out to the events. We extract the signal from the noise, a lot of signal at this event, really around a number of things, but specifically around storage. Storage has been a pain point in the VMware community for a long time. Basically the better part of a decade. And Rich Peterson is here from Sandisk. We're going to talk about some of those problems, what Sandisk is doing. Rich, great to see you again. Thanks for coming on. Thanks for having me. So, you guys have been through quite a massive transformation. Some really strategic acquisitions. The market is changing. And we think you're really well positioned. Set it up. Talk about the business where you guys are at. Well, like you say, Sandisk is a well-known leader in the flash industry. Everything from fabrication to the retail presence over 25 years in the business focused exclusively on this technology. But over the past five years plus, we've been expanding our focus to cover the enterprise space. With, as you mentioned, acquisitions, including client technologies, Fusion IO, FlashSoft, where I came from, and others. And we really do see the opportunity to address the storage latency issues in virtualized computing with flash technologies at every level. At the storage, in the server, and in collaboration with software technologies like VirtualSand and software technologies that we bring to market like the FlashSoft software. So what do you have going on here at the event? Well, this is probably the biggest VM world for us yet with respect to FlashSoft because this year we announced that as a result of our collaboration with VMware, we're going to be introducing the next version of our FlashSoft software as a virtual data service that's enabled through a new set of APIs called the vSphere APIs for IO filtering. We were actually a VMware's design partner in the development of the caching capabilities of that framework. So we've been working with them behind the scenes for about a year and a half now to ensure that we can bring this technology to market and it can be delivered and certified VMware ready. So you've been talking about this, David? Yes. You've been telling me it's a big deal. Why is it such a big deal? Why are you so enthusiastic about it? Well, because of the bottleneck of IO within the VMware architecture. And this is the first time they've done a previous job on it. What was it called? The VMware technology? Yes, the VMware. Yeah, the VFlash read cache. That's right. They had some stuff in there but really it wasn't that functional. It wasn't really getting over the basic problem. So it limited your ability to really exploit the... What they've done is introduce you and others into a proper architecture and essentially allow the whole of the stack to be bypassed so that you can go from A to B without going through all the layers. So you're cutting the line in the horrible storage stack? Yeah. Fair way to say it. That's not a bad thing to say. You know, VMware had a number of objectives with this framework. The first objective obviously was to enable write back caching as well as write through caching so that we can accelerate an application's write operations as well as their read operations. This is really important when you're gonna be running a transactional database or a business intelligence application in a virtualized environment because those types of workloads have a large number of random small writes and that's what's really frustrating for storage. But beyond that, they also had the objective to establish a framework that would enable third party data service or software providers to provide caching solutions that would be integrated in an authenticated in a good way with ESX. This means getting third party technologies out of the kernel space and integrating and intercepting IO in the user space. That was a huge objective for VMware and they put some of their top engineers on this framework development. It's been a real privilege for us to work with them. It's really, as you and the other one is, what is it? Primary IO, isn't it? Who's the other partner that you had with you? Well, so in the framework development, EMC was the partner for the replication filter and we were the partner for the development of the caching filter. Okay, so you guys both worked sort of independently with VMware or was it a sort of? The filters actually are sort of at separate layers of this framework and that's for good reason. Simply put, when you have the replication filter at a higher altitude than the caching filter, a write back cache is not gonna interfere with replication. So that's all good. And so there's been a lot of thought about enabling write back caching in the vSphere environment yet making sure that you take care of any problems that could occur such as ensuring the coherence of snapshots or ensuring that write data that's cached is replicated for high availability in the event of any kind of event in the cluster. Or vMotion. vMotion, same thing. Yeah, yeah. So there's a lot of, as the design partner, I like to joke that we have a number of Russian engineers on the team. And if you want somebody that can think of everything that could possibly go wrong, find a Russian engineer. But the truth is that we did have to consider all of these failure scenarios and make sure that we had proper protection of the data. Because again, write back caching as a new, it's a new idea for a lot of administrators. And if you're gonna bring technology to market that is certified VMware ready, that has the assurance that we're gonna take care of you and your data, you've gotta get it right. So how are you going to market with this? I mean, this is a new area for you, isn't it? Because this is not just your technology, it's anybody's technology of support. That's true. The objective of having an API is to create a level playing field for your ecosystem partners. I think we do have an advantage in having been the design partner and working on this problem for a longer period of time and in a greater level of detail. But I think the go-to-market strategy for SanDisk as a whole is a bit of a hybrid strategy. We do have a direct sales effort. We do have a channel operation, but we have a very significant OEM presence as well. And this comes obviously from our hardware business in which we've been supplying solid-state technology, SaaS, SATA, PCIe, even UltraDim through to storage and to server vendors. So we do have a very strong reseller relationships already in place with Dell, with Lenovo, with HP, and with HDS. So our go-to-market strategy is very much in collaboration with VMware as a technology partner, but then also with these major server vendors as business partners. Excellent. So that's a very different way of thinking about software in this particular space, isn't it? Not necessarily for you, but most software vendors are not thinking of it as an OEM player. I think it's a solution-oriented approach. We need to work with the decision maker that's looking at what the architecture for their VMware cluster should be and to understand their options for using Flash Include, not just Flash in storage, but Flash in the server as an intermediate cache that will enable them to really prolong the useful life of the existing storage infrastructure that they've already got. So I got a question from the text. I want to do it on the crowd chat. So the question is, SanDisk strategy, the market's changing, all kinds of startups popping up all over the place, but SanDisk is in a unique place in the value chain, let's call it. So what's the strategy and objective for growing the company specifically in the storage space? Well, clearly, we see the enterprise as an opportunity where we've only seen the tip of the iceberg for Flash. We could talk about the other systems that we've been developing, like the Infiniflash systems and the Ion Accelerator and many of the other technologies that we're bringing to market. But as you mentioned, we are in an interesting position in the industry in that we are a supplier of technology and capability to the vendors that the end users have had long relationships with over decades. And so we want to enable those vendors to be successful with their customers. If we can do that, then I think the future for SanDisk as an enterprise Flash solutions provider is very bright. Because you've got an interesting stack, you've got a lot of pieces. And so essentially the strategy is to build solutions for your partners to take to market. Right, right. And so that's cool, sounds good. The interesting thing is that you can potentially do this for so much less than the traditional storage model, so much less expensive than the traditional storage model. And does that threaten some partners potentially? You know, to some degree, because I'll give you a simple example, people ask me, well, if you have your caching technology which works with existing storage arrays, so you're doing server Flash, but saying keep the storage array and then your colleague over here is telling me about all the virtual San, all Flash virtual San configurations that you're validating and supporting aren't you too competitive. And the simple matter is that, yes, these are alternatives, but the decision fundamentally lies with the storage architect, right? They're going to map their roadmap. They're going to say this is where we want to go. It's really incumbent on us to make sure that whatever choice they make, and in our case with Flashsoft, it's server Flash over your existing storage infrastructure. Now, you may be spending less on that storage infrastructure. You may be enabling storage functionality at the host level now, instead of at the array level, but that's sort of the natural evolution of our industry. It is, and there are a lot of good reasons to do that. There weren't maybe 15 years ago, you know, wouldn't work as well, and we pushed data out for a good reason, but it's coming back for a good reason. And so I guess your point is, we have to drive technology, we have to drive costs down, that's what this industry requires. So, damn the torpedoes, we're doing it. Very much so. I mean, the technology that we're talking about today is one that we began thinking about seven years ago, when we began to imagine what would the world be like when you could buy a terabyte Flash device and put it in, and it sounded crazy at the time, right? But we knew that was coming, and what we have to be thinking about as software developers today is what the world's going to look like five years from now with latencies that are 100 times less and capacities that are 10 times greater. I mean, when you look at the storage industry historically, it's been taking a Seagate or whatever disk drive and adding a lot of functionality to make this actually work as it was a very, very difficult piece of technology to make it work. So, and the markup on it in terms of times the original cost of the disk is 10, 15, 20 times more than the disk. With Flash, that's not nearly the same markup, is there? It appears that there are a lot of problems in terms of the fundamentals of dealing with the Flash and ware-leveling and performance, et cetera, have to be done by the Flash manufacturer themselves because it's unique. I think you're right about that, and certainly our team is extremely strong in that area. Technology's to get optimal utilization and performance from the Flash resource, but at the same time, the software that we provide also has to use Flash resource intelligently. We can't treat, when we're talking about a caching capability, you can't treat solid state storage like spinning disk, nor can you treat it like a DRAM. You know, there are certain capabilities in our software that were built specifically because we knew we would be using non-volatile NAND Flash as the caching resource. And so our software is designed specifically for that, for the physics of that, you know? So following up on what Dave was talking about, is there an opportunity, though, for you to own pretty well a complete stack from, because you'll be able to optimize the software in a whole number of different places. The way that end users are using Flash is so different than the way that they were using disk. Right, right. I can't comment about whether Sandisk has a master plan to own the entire stack, because from my perspective, a lot of our success really does hinge on empowering the users to get the technology that they want the most. A simple example is the fact that our software will work with any vendor's Flash hardware. This is simply because, when you're working in partner relationships with server vendors, the requirement for a tightly coupled hardware software opportunity would lock you out of any long-term relationship with that vendor. Right, right. Now, we obviously hope that their success with our software will lead them to consider our hardware more, but again, we have to win on both fronts. Well, but these are, as the industry evolves, as you evolve, there are opportunities that you're going to assess based on customer demand, and you see opportunities, it's pretty clear you guys aren't afraid to go after them, and you came to Sandisk as a result of one of those opportunities, and it's, to us, it's transformed the company in this space, and it's really exciting to watch. We came to Sandisk because at Sandisk, we can engage in a multi-year, behind the scenes, partner relationship in development of a new technology with a partner like VMware. It'd be very difficult for a startup to do that. Deep, deep partnerships, integration, we see Amazon coming to one end, we see folks like Sandisk having tremendous opportunities, so really appreciate your rich coming back into CUBE, and good to see you again. Thank you very much. All right, keep right there, everybody. We'll be back with our next guest right after this. This is theCUBE, we're live from Moscone, at VMworld 2015.