 Hi, we're back. This is Dave Vellante. We're live from VMworld 2011. SiliconANGLE's continuous coverage, siliconangle.com, siliconangle.tv, cloud.com, sorry, cloud angle, services angle, Wikibon.org, check it out. We've got a great panel here. We're going to talk about storage optimization. We have a CUBE alum. Actually, you guys are all CUBE alum. Noemi Gresdorff from Cambridge Computer Technologies. Welcome. Thanks for coming out. You made it through the storm, okay? Yes. And Greg Schultz, Storage I.O., a famous blogger, and Mark Farley, who was... What are you now? You're no longer three-par Farley. No, I'm HP Sisyphus. HP Sisyphus, that's right. Beautiful. And famous blogger, prolific blogger. So we're going to talk about storage virtualization. First of all, welcome back to the CUBE, VMworld 2011. I wonder if I could just get your initial take. Noemi, maybe it's not fair for you. Did you get out here before Monday or no? Were you delayed by the hurricane? No, I got in last night, late last night. You were delayed by the hurricane? So I'll start with Mark. We talked about this on the last panel, but some of the interesting stuff that you've seen at the show, and Greg's got a really interesting storage innovation that he's going to talk about, but what have you seen that really has blown your dress up, so to speak? Well, the thing that's most intriguing to me that I said before is this whole... I'm trying to remember it, VX LAN. That's interesting. That's going to be some new thing. There's going to be a lot of discussion and questions about it. There's one thing I like, it's mystery, right? So to me, that's the most mysterious thing going on here. Otherwise, I think the enhancements in vSphere 5 to storage are interesting. What's coming with that? You know, VASA, what happened with VAI, what's happening with SRDS, that kind of stuff. I think it's really interesting. The direction that SRDS will eventually go, I think will be interesting to watch. With all these acronyms, we know we'll have a job for a good time. Now, Greg, you're talking about some very interesting concepts in storage, right? User self-provisioning? Is that what I understand? Oh, there's absolutely self-provisioning, all kinds of fun things happening. And as Mark mentioned, lots of interesting things happen around vSphere 5. Yeah, so I think I saw that same, you know, do-it-yourself storage out there on the Boulevard. Oh, absolutely. You know, they got this new show. Maybe it is competing with yours called Storage Wars. Yeah, you've seen that where they auction off the storage bin and you get to go in there and figure out what it is. It sounds like a threat. Maybe we're going to sit now with data auctions. We've written about the Storage Wars, but it's a whole new perspective. Oh, yeah. A whole new definition to big data. So, Noemi, have you had time to check out the floor? Have you seen anything interesting? What are you looking for at the event? Well, Cambridge Computer Services is a national storage reseller, a value-added reseller and integrator, so my responsibility is really to go out and understand what is going on in the industry, what are the interesting technologies, innovative products, and figure out which ones our customers could benefit from the most. So, I've been walking the floor and checking out the different... So, Noemi, let me start with you. So, you are an expert in this field, actually a former expert, because you're an IDC analyst, every IDC analyst is an expert. Oh, IDC gardener said it must be true. I used to work at IDC, I can make those kind of jokes. But no, seriously, you have a great knowledge in the space of storage optimization. And I think one of the hard parts about your job when you were there was, what is it? Because you had to size it and define it. What is storage optimization to you? So, I actually had this conversation with somebody walking through the exhibit floor. What are the customers are looking for? And one of the things that they're looking for is obviously storage optimization. But I think it has two very distinct meanings. The first is they want to optimize for performance, which means that they want to get the maximum performance out of the smallest footprint that they can. And I think that a lot of the innovation around the use of SSDs in storage systems or putting SSDs in the host, using them with hypervisors, there's a lot of innovation, a lot of interesting architectures that are coming out around that space to optimize performance of storage systems. And then the other side is capacity optimization. How can I store as much data as possible securely and safely in the smallest amount of data footprint? So there's sort of two different tracks that we can talk about in terms of storage optimization. Greg, you always have an interesting perspective on things. What are you seeing in the storage optimization world? Build on what Naomi just mentioned here, which is it's about time, which is performance. It's about space, which is capacity. What's been interesting, Naomi, is the last several years a lot of the discussions, 90% are focused probably around 10% of the problem, maybe around backup, as opposed to moving further upstream to where the source is. In other words, looking at other aspects of data footprint reduction, archiving, backup modernization, compress, consolidate, data management, all those things that if you don't check them up in front, further upstream, they just keep trickling down. You can take it even one step further, and you can say that as the disk drives are getting bigger and we have greater density, there have been some concerns around the use of RAID, and there's this conversation around using erasure encoding, and there's a couple of different applications. Because of the rebuild times, it takes like eight months to rebuild a disk drive. RAID 6 can be a couple of weeks long. The idea that where object storage comes into play and then further even the dispersal algorithms that we're seeing being productized, that's a form of optimization if you think about it moving up the stack. So RAID 6 takes maybe 25% overhead in terms of capacity. But then if you want to have a DR set, you have to replicate it. That's double that. So you've just doubled the capacity that you're consuming per site with RAID 6 overhead. With the dispersal algorithms, for example, you can gain higher redundancy without having the RAID rebuilds while doing a geographically dispersed with maybe 1.6, 1.5 overhead versus two and a half times overhead in capacity. So there's a lot of different technologies at different levels of the system infrastructure in the software layer that add to the whole concept of capacity optimization as well as performance optimization. But moving up the stack, right to left, what's your perspective on this, Mark? I mean, your scope is maybe a little different. What is storage optimization? Maybe even put in storage efficiency to that whole mix. What does that mean to you? Yeah, so I think there's three product classes for storage efficiency. And I'm talking about capacity now, not so much performance, but their reclamation, dedupe, and compression. And if you apply those technologies in different ways, they can fit for different types of applications. Certainly dedupe is a fantastic technology that we use for archiving and backing. And NetApp uses it for primary storage. Nobody else is really doing that yet. That's an interesting area. It'll be interesting to see how that grows or how that develops. Reclamation was a hot button about a year ago. There was a lot of stuff going on last summer, not the one that we're in, but the summer before where there was a lot of talk about reclamation. It's kind of cooled off, but I think it's going to come back as there's this notion of virtualization sprawl. It's so easy to create virtual machines. It's possible to create data repositories out there that then get forgotten and unused. And you need tools to try to reclaim that stuff. Well, first you have to identify it, and then you have to reclaim it. And so you're seeing software vendors like VMware, like Microsoft, coming up with APIs that help storage arrays reclaim space. And turn it back to the free pool. That's an interesting area. I think you're seeing a lot of development there. But I just think people are going to find ways to get the storage back that they already paid for instead of having to go out and get storage with POs. All right, guys. I'm sorry that we're so tight on time here, but we've got to wrap. I'm actually going to ask you guys, you know, ooh, ooh. We're done? We're done? Oh, no, no. I wouldn't have gone on and on and on like that, but I know we're running out of time. No, that's okay. We've got a little bit of time. But we've got to wrap. So I want to finish with one last question here. All right, let's put our binoculars on or telescopes if you want. What's next? I mean, we've seen DDo, we've seen, we're talking about compression, reclamation. What's next in storage optimization, storage efficiency? No, I mean, what do you say? Well, like I said, I think that there's a need for other forms of optimization in terms of capacity where you're dealing with achieving greater redundancy and safeguarding your data without the penalty of multiple, multiple, multiple copies. And then there's, I think there's a lot to be said for storage optimization in the performance space and there's a lot of activities. Real time. Real time. Yeah. Great. Mr. Schultz. Putting it all together, leveraging what you currently have, leveraging the new technologies, in other words, the ABCDs, A, Archive, Backup, Backup Modernization, Compress, Consolidate, Data Management, DDo, Getting into some of the others. The storage tiering, space-saving snapshots, Reclamation, going off and finding that orphan storage that's just sitting around taking up space that you forgot about a VM or something that you'd provisioned on it, but putting all those different pieces together but focusing across everything, not just backup, online primary, are you seeing the right tool for the task at hand? Standing. Great. Mark? I think the application of DDoop in wide area environments, Noemi had touched on this a little bit. The idea that said, I can DDoop something remotely. I don't have to send it, but I can still rehydrate it at another location if I need to. So you share metadata instead of actually moving both amounts of data and then re-DDooping them. This notion of speaking the same language across the portfolio and not having to rehydrate it. Good vision. All right, guys, listen. Thank you very much, Noemi Gresdorff, Greg Schultz, Mark Farley. This is the Storage Optimization Spotlight at VMworldLive. Dave Vellante of Wikibon.org. This is SiliconANGLE's continuous coverage of VMworld, and we will be right back.