 Okay, we're back, this is Dave Vellante. We're live, VMworld 2011. I'm here with Milan Shetty, who is the CTO of HP Storage and a cloud advisor, former cloud advisor. So welcome. Thank you. This is the spotlight that we've called Beyond Storage Virtualization. And Milan, we were just talking to Savvis. You probably heard some of the conversation about some of the cloud service provider requirements. We talked at the top of the segment about storage virtualization and how it really changed the mindset around utilization. So I want to have a discussion with you about that, but I want to start with, actually let's start with your role. So you're coming from a position of cloud advisor into CTO of HP Storage. Talk about cloud advisor and talk about your priorities as CTO. Absolutely. So I actually came to HP a couple of years ago via the acquisition of Ibrics. Ibrics, I was the CEO of Ibrics back then. So I've been dealing with the cloud environment for a while and the core mission of HP Storage, as we see is helping and leading from a vendor standpoint to assist customers making the transition from enterprise to cloud. And that is in one sentence if we want to call it out of the core mission, that's our core mission. So if you look at the acquisitions HP Storage has made in the recent past as well as where the products are going and what we have announced this week and last week between you referred to Savvis, it's all been around helping through the transition. As you've seen from the three-part standpoint, we had the pioneer of the thin provisioning storage, multi-tenancy, all of the components which are now considered core and bread and butter to having the enterprise go to the cloud including the notion which we have now introduced around the federation and peer motion with CMAC. I know talked about just a few minutes ago. What we are trying to do is how do we extend that not only with three-part, we also have the left-hand products which is a product which sits right below the three-part in our product portfolio also have the peer motion sort of a knowledge to make the federation as well as the data mobility features available. So it's really the mission of how do we take the enterprises to cloud and make that transition easier is what we're core and primarily focused on. Yeah, so you've got a much more diverse portfolio than when you were at Ibrics. Obviously that's a challenge. We heard from CMAC that peer motion is peer-to-peer. It's not, your strategy seems to be not to put in a virtualization layer. You seem to be actually very much averse to that. Is that true or am I reading too much into that? Not quite, if we take a step back, not quite true. So what we're doing is that I would call that peer motion is just an introduction. Federation is just an introduction in where we think we are going to go further in our product portfolio. Where we see is, so if you look at HP as a company, so we've got the servers, we've got the storage. We do a lot of data mobility, federation, peer-to-peer motion components now within a three-part array or between three-part arrays. Where we expect these things to go is active and very clever cache collaboration between elements which exist in the storage but also something like SSDs which exist in the server. So you can expect us, so in a way I visited it to ask in just pure virtualization because it goes beyond storage virtualization. It is how do we make use of all of the data mobility components and the features we have in the three-parts and left-hands and make it expand over to storage which exist in the server domain. So if you think about the world where there are blades and there are lots of SSD drives in them and then they are attached to three-part or left-hand, is that SSD and the storage which also exist in the server now becoming eventually the part of the federations and the peer-to-peers is, we believe, something which is going to be extremely critical in where the market's going. So it's way beyond storage virtualization. It's actually cache virtualization or call it solid state disk virtualization to take advantage of the data and the application co-location because all this, how do we get this cache and how do we get the data and how do we get these data mobility features closer to the app, even beyond the storage right into the app layer itself and that's what we're going to be. Interesting, so talk about your vision for cloud and cloud storage. So how does all this fit in, draw on your cloud advisor background and paint a picture for us as to what storage looks like in the future? Sure, so if we look at, we probably have to take a look back at the requirements. So virtualization was a very disruptive technology which made applications very mobile. It broke the boundary between the physical and the virtual. The networking infrastructure now exists today that people can host their data in the network and not have to worry about in a lot of cases latency and everything because there are technologies available even to speed up the things in the networking. From a cloud standpoint, I think if there is one area of virtualization where virtualization had been traditionally weak, it was actually in the storage area, right? People were still dealing with the physicals and not much data mobility. So the first installment of making enterprise applications cloud aware was really in picture of a whole bunch of storage devices from the federations and the peer-to-peer motion which we introduced in three part and multi-tenancy and thin provisioning. Those components are very integral to making sure that your data is available to your application and to VMotion as VMotion or the virtualization as it's moving. The data is also available and data is also moving with it. So I think we have provided that level of mobility now into the frame, into the storage infrastructure. The next step really is how do we extend that back to my original point around into the app layer itself and also the cache which is available to the app. So the way the world would be then down the line is whether as the app is moving, the data does not have to move. Only the hard data has to be accessible into the local cache or the local solid state devices of the blades or the servers where the app is sitting on. So I think once we do that, we are really breaking the boundaries of network latency also and making sure that the hard data is available closer to the application itself. So talk about what this means for organizations. Obviously it's much, much more simplified to manage. Are we talking about other business value in terms of application performance as well? Absolutely, application I think from a business value standpoint, the first impact which our customers are going to see is that applications are going to be running much and much faster. I'm not talking about latency which could be in the milliseconds or even microseconds. So all the enterprise applications which have struggled to get onto the cloud because they didn't quite handle the latency which existed in the traditional cloud environment I would say, are now going to be able to expedite faster because the barrier of latency has gone. Hard data is available in the local premise. Not only in the local premise, it's actually available in your server complex. And extending the data tiering capabilities, extending the federation capabilities from the storage to the server tier and the server caches is going to adapt all of these bunch of enterprise applications which didn't go to cloud, always were hesitant to go to cloud. You'll also start going to cloud. If you look at about it, undoubtedly it's going to come down to application performance getting faster is the single most barrier for all of the enterprise or almost all of the enterprises going to cloud and we'll do that. If you look at the cloud applications are the first set of applications which are in the cloud are backup, archiving or data which is right once, read many kind of applications where the latency has never been a factor. That world is going to change. I expect that down the line once we achieve that goal through our product lines, we can actually very easily, I can envision seeing oracles also in the cloud. Yeah, I mean it's all the applications with crappy performance today that are in the cloud. Exactly. And so if that's it, the cloud is not going to live up to its hype, is it? That's correct. Okay, that's a great vision and I've not heard it articulated that way before. The other, so you're sort of describing the archiving backup as sort of cloud 1.0 applications and the new emerging applications. Certainly the traditional block based stuff, oracle. Big data is another one you're seeing now. You talk, you know, Ibrics, that's big data. That's correct. Now you're talking about big, fast data as sort of the next wave of these applications. Absolutely. I mean, and you could think about, right, oracle and the traditional structured data aside, even in the unstructured data and the big data to your point, the business analytics which has to be run on subset of the metadata cannot be in the cloud today because it's low from a latency standpoint. Now, if that metadata can be in the server complex and the storage pushes it up because its intelligence has got all the heuristics and it knows which data is hot and just pops it up into the cache layer or make sure it never leaves the cache layer, that is huge. That is absolutely huge. So, based on what you're saying, customers should be thinking about metadata management. Correct. And maybe thinking about it, first of all, thinking about it because a lot of customers don't think about metadata management and the ones who do may be thinking about it in a different way. What would you say to that? What would you advise them there? Right, I think the question generally comes around, I do see a lot of customers, they don't ask the question about, I want the metadata performance to be better, but they ask this question about, I have this solid state device which a lot of vendors are pitching and there seems to be a trend around SSDs everywhere. What do I do with it? Should I be worried about it? So, I think the question has not, sometimes it takes a while for these questions to really set them down. They've seen this media, a fast media which can be put in the servers or in the storage. How can they benefit from that? And I think from a community standpoint, a community such as viewers in Wikibon and the blogging community, we can educate the customers around, you know, one way you can really make use of your, this new trend happening in the faster media, someone once said it, I can't take credit for this code, is that SSD is going to be the new disk and disk is going to be the new tape, right? So, what does it do to customers? And I think if we drive down to the customer education and the market education as well as the community education in the storage industry and you know where you can really take advantage of this is to make your metadata access faster, metadata read faster and which will hence result in your apps running 10 times, 15 times, 20 times faster and these numbers are not, just numbers when I'm making it up. I think with SSD and metadata, all the stuff coming from SSDs rather than coming from spinning media, we're talking magnets of orders of revolutionary change. One to two orders of magnitude, yeah. So this, I like the vision you're laying out, it's a much more than bigger, faster, better. You know, we're talking about a complete transformation of the IT architecture which has organizational implications and orders of magnitude potentially more performance and then that's going to drive tremendous business value. Milan Shetty, thanks very much for coming on theCUBE and sharing your knowledge and your insights. We'd love to have you back anytime and it was a real pleasure talking to you. Likewise here Dave, great to be here and look forward to talking to you again soon. Okay.