 Everybody, we're back. This is Dave Vellante. We're live, day two from VMworld 2012. We're here in San Francisco with the Moscone Center North in the hang space. So stop by and see us. Check it out. Stop by theCUBE. Grab a beverage. And I'm here with Stu Miniman, my co-host. And we're joined by Laz Veccherides, who is with Dell. Long time CUBE guest, CUBE alone. Welcome back. Good to see you again. Thank you. Thanks for having me again. So we're here again at VMworld. It's the perpetual party here and all kinds of innovation. We're hearing about software-defined, networking software-defined storage. We just came off the Dell Storage Forum in June. So why don't you give us the update? What's been going on since then? And what do you guys get going on here? We actually outed a couple of interesting concepts that we've been hiding for quite some time at Dell Storage Forum. And I reiterated them again today and yesterday. You mean we didn't talk about those in the CUBE last time? No, it's a matter of fact we did. Okay, I thought so. I think it was Stu. We have this concept which we called host-virtualized storage, which it looks like is exactly the same as software-defined storage. Effectively, what we're doing is we're defining what the storage IO platform is. It's a software workload that runs our firmware, the Ecologic firmware, and allows an immense amount of flexibility as to how you define your storage services. So as we continue to integrate in virtualized environments, our customers are going to be able to control these policies for their compute workloads at a much finer level of granularity. You'll be able to define a software workload which is your storage. You'll be able to define where it lives, how it behaves, and you'll be able to define it on a per-application basis. And you can do it many, many times. You don't have to go out and buy a physical array for every application. And so it completely changes the dynamics of the data center and how you pay for things and how you configure things. The CAPEX cycle is a much smoother thing as a result of this. So it's quite revolutionary. Let's talk about how we got here. So years ago, Merit's called it the software mainframe and others have called it the cloud operating system. Now the new buzzword is software-defined networking. So really what we're talking about here is VMware uses the terms of abstract, pool, and automate. Sure. Okay, so that's the sort of high-level umbrella. So then, and they've done clearly a good job of that compute and memory storage has been hard. Sure. In fact, the ecosystem had to do that, right? Sure, sure. Integrating APIs and the like. So where are we in that whole vision of making storage invisible? It sounds like you're saying we're there today. Well, not quite. I think there's a lot of work to be done. If you listen to the keynotes this week, you've probably heard a little bit about what they're calling virtual volumes or V-Volves. I guess it's granular or VM granular storage. VM granular storage, right? That is a huge initiative that we're participating in along with all the usual suspects. That is, I think, the ultimate vision for where we want to be. Where you get to the point where each VM has its own storage policy. Up until now, all the storage vendors have had to deal with a per-lun policy. So you put a whole bunch of virtual machines on a line and by virtue of the fact that they share that line, they share the policies, they share the performance. And it's very hard to police that in a large data center. So what we're doing now with V-Volves is basically breaking that out. Now, that's the ultimate vision. And on top of that, that kind of policy-based automation, you probably want a whole bunch of other services. And this is where the host-virtualized storage concept comes in because there you have other policies like snapshotting that might be application-specific. How many snapshots, what kind of replication, and caching where you want to place things. Those can actually be encapsulated in a piece of software. I wonder if we can dig a little deep in there. You talked about caching and tiering and where things fit. So Wikibon CTO David Floyer says what we really need is an IO-centric model of architecture for the data center. So while we know flash is this kind of revolutionary wave coming on, it's not just about kind of putting it into our existing technologies, but how do we redesign and re-architect that? Can you talk about what Dell's doing in the flash space? Sure, sure. Another thing that we talked about yesterday in our breakout session is the RNA technology, which we're going to christen fluid cache. And we have a couple of interesting innovations that are going to be coming out over the course of the next year. Our 12G servers have these front-loaded NVM cards, which they look just like disks. You put them in, they're 175 to 350 gig of just pure flash PCIE attached. So they're much, much faster. And you treat them like disks in a disk array. So this is going to be a very, very interesting thing to have on every server in your data center. So effectively what we're doing is taking the RNA technology and building a fluid cache, which is a distributed caching substrate that lives right next to the compute. And it behaves just like the ecologic storage in the sense that you can add and remove, vacate, upgrade. It just expands and contracts dynamically. So whenever you provision an application, you can provision a certain amount of flash on the local storage for it as well. And that will actually follow the application around because, again, we're distributed storage infrastructure even at the front end. And we have all sorts of really interesting technology that we're developing to coordinate the data protection mechanisms in the back end. So you still do need to have a SAN to protect your data. And snapshotting and replication still need to be done there. And there is a delicate dance that we have to coordinate in order to make that all possible. One of the key things that we're doing that I don't see a lot of other people doing is building a full read-write flash infrastructure on the server. So it's not just writing through to the SAN. We're actually taking the writes on the server or making the server flash highly available. And what this will do, if you look at what the vision will look like over time, what you'll see is that you'll have some really, really, really fast storage very close to the compute. And you're going to have much slower stuff, sort of in the center of your data center. And it's doing a lot of the data protection, the backup, the replication, those types of things will be relegated there. And so, you're kind of seeing storage in the data center sort of bifurcating into two buckets. And one bucket is really, really, really fast and looks like memory. Another is more economical and has a lot of other management attributes that you need for retention. And so, yeah, we actually might even, there might even be three or four different buckets here. So one of the follow-up questions I want to ask you on this is when virtualization first came in, it really broke disk. It took us many years in the storage industry to kind of take the lones and how we manage this environment and change it. So have we learned from that? Flash is kind of this big change. And some of the first flash solutions that came out, oh, it's not really going to work with Vmotion. And now it seems we're saying, so what do you see the ramp up in the virtualization environment for flash? How are you guys seeing that progress? Well, you know, we have to work with VMware and all the other platform vendors in order to advance the cause of flash, right? And Dell were uniquely positioned because we actually sell the servers as well. And we are spending a lot of time trying to push our vision with the people that are actually building the operating system. So I can't talk too much about what we're doing with VMware and their competitors up north. But they're very much aware of what needs to happen and what the evolutions are gonna be on the server platform. And over the course of the next year or two, you're gonna start to see some really interesting innovations with how we deliver value of flash in these operating systems. Like I said, we're gonna virtualize it the same way we virtualized disk today. And what we're hoping to do is make it so fluid that an application can take advantage of flash like it takes advantage of memory today. And we will police it. We hope that the operating systems and the system software can actually have policy infrastructure and things underneath that to in real time determine who can have the flash and who really should be talking to something slower. So, okay, sorry. So I really want to follow up on that. And so in your view, will the point of control for that decision-making be the server, right? Won't be the slow storage, right? Will it? Well, I think there's a real interesting discussion about, well, who actually owns the policy engine? And- Can slow storage own that policy engine for fast flash up at the server load? I think there needs to be a bigger brain somewhere outside of the slow storage and the fast cache because each one of those needs to make local decisions that are based on information that's only available locally. But you have a policy engine somewhere in every data center that really is sort of the control brain of all of these things. So, there really isn't anything out there right now. New model, as you're proposing, yeah. And I think that that's an opportunity that you're going to see a lot of people sort of discussing, I bet you this is something that will come up next year as we revisit these subjects. Do you agree of it? Stu mentioned David Floyd has also made the statement that all active data in the near-term, near to mid-term, going to be in flash. You believe that? Yeah, I think that's an outgrowth of what I said, active data will end up getting, if you look at the fluid cache architecture, active data will get pulled into that server. And the data that's not frequently accessed will naturally get pushed back out. And you would not, ideally, never access it on spinning disk, right? Wouldn't that be the perfect world? Well, that would be a waste of power if you put it in, but yes, it would be really easy to meet the SLA if you never actually purchased or read it. Well, I'm saying, if I access it, I want it to come out of flash if it's active, right? Yes, yeah, I do agree with that. And you, you know, what we are finding, actually, I've seen a lot of surveys, I'm sure you've seen the same kind of data, about 20% of your data is reasonably frequently accessed and 80% of it, yeah, 80% of it is, fine, we can agree on that. Yeah, the Pareto rule, right? Yeah, the vast majority of it's going to be sitting in the bit bucket. Exactly. So, Laz, you've got a long relationship with VMware on the engineering side, you know, going back before Dell, so can you speak to kind of the ecosystem, a partnership that you've got, you know, a lot of people talking about the kind of the relationship of VMware and EMC, and you know, personally, your engineering group, how's the relationship between Dell and VMware? Have you seen a change over the last, you know, six months or a year, you know, where, where are things? That's a good question. I think the VMware people are incredibly gracious with my team, and I have so many touch points into VMware with various parts of the engineering organization, with various parts of the product management organization, they have always been very, very open to conversations, and I don't think I've seen any real changes over the past year or two, you know, they've become a much larger organization, and so that happens, they're just more people, but then again, my organization's much larger, and we just have more touch points, and it's a matter of coordinating everything and making sure that everyone's singing from, you know, the same hymnal, but I don't think that there's been a change in the openness, I don't think that there's been a change in the depth of cooperation. We are involved, as I mentioned in all the big initiatives that the VMware has, you know, undergoing right now, and some of the things that were spoken about yesterday and today, and the keynotes, you know, I know that Dell is always being mentioned, we are, especially in the equal logic side, always there, talking about virtual volumes and, you know, VStorage and the like, and so we plan to continue to do that. You know, there is an interesting question about whether this has become a big voice club, and, you know, that's a better question to ask. I think, you know, I was lucky because, you know, I started talking to VMware when we were equal logic and we were very, very small, and VMware wasn't all that big either, and it was a different environment. Right now, you have the big four, or five, depending, I don't want to offend anyone, but, you know, there are a lot of startups that would like to be equal logic again, and I'm not sure that it's easy, just because of the nature of the beast, there's so many places to talk to VMware. Yeah, and it was not easy for a small company back then to get VMware's attention, they didn't have a ton of storage resources, so. All right, Laz, well thanks again for coming on theCUBE. Always a pleasure, and good luck for the rest of the event. Okay. See you back here. Thanks a lot, spring Michael, bye next time. All right, keep it right there, we'll be right back with Frank Slutman, Tech Rockstar, CEO of ServiceNow, former CEO at Data Domain, is up next. Keep it right there, we'll be right back. SiliconANGLE.tv is theCUBE.