 Okay, we're back. This is Dave Vellante of Wikibon.org and we're live here. Silicon Angles coverage of VMworld 2012. We're here with our good friend Jason Buffington, a multi-time CUBE guest. Jason, welcome back to theCUBE. Thanks for having me. It's a pleasure to see you. We love your insights. ESG, you guys are doing a great job. Really growing like crazy. Every time I see somebody from ESG, you got more guys there, more research. You guys are pumping it out like crazy. It's fantastic, so congratulations on that. We're here at the spotlight on VMware Backup and Recovery, an area that you know well. Well, first of all, what's your take on the event here? What do you see, what's interesting, and particularly from a data protection standpoint? So this is actually my first VMworld. I've been to all the other events this season, but this was actually my first time to come to VMworld, and it's a pretty neat event. I've had a really good time. It is. It is. Certainly I think my favorite announcement of the day was with vSphere 5.1 and the introduction of VDP, the data protector product that's now built in. That's a really big deal, and I'm surprised that in fact the only time they saw it on the keynote part was when we were doing the demo Palooza competitions and Chad nailed it from a demo perspective, but there's some good stuff in there. Yeah, well, I mean, it's always been a challenge, as you know, of VMware data protection. I mean, I always go back to the days of VCB and finally VADP had the ecosystem to solve that problem and VDR, you know, it was a good effort, but really just didn't get there. So I think finally, I mean, Chad was on theCUBE talking about the efforts and Stephen Manley and Guy Churchill as well, talking about 60 engineers. It was not a trivial task. No. But so what do you make of that? I mean, so now it's embedded, it's free, so it's going to potentially eat into some of the high end. Yeah. But that's a bold move. What do you make of it? So a couple of things. So I like how you set that up. There's, you know, we had VCB, the 1.0, that just min bar, let's get something in the box so you can back it up. And, you know, coming back from my old history, I know that when you develop a platform, if you're going to put all your eggs in a basket and you can't protect the basket, no one's buying it. So VCB had to be out in there out of the box, but it was days that we'd like to forget, right? And then there's VDR, and with a lot of things in technology, it really takes to about a 3.0 kind of perspective before you really kind of get it right. And VDP looks like it's going to do it right. It was not a trivial effort, right? I think the thing which was really interesting to me is, you know, typically when you look at a built-in something into a platform, the built-in's kind of wimpy, and you know, you're going to have to still do something else for almost everybody. It's muted, right? Yeah, yeah. So this is built on Avamar code, right? So this is not untested, this is not immature stuff. This is built by a product that knows how to back up a virtualized environment. That's step number one. And then certainly, you know, we did some research earlier this year, and I'm hoping we get a chance to kind of talk about that a little bit more, but one of the big trends that we saw there was DDoop is a must. Back up to disk without DDoop is non-viable, particularly when you're talking about virtualized environments. And so the fact that DDoop is in there, that you still have granular file level recovery, I mean, this is not a wholly neutered thing. It's certainly not Avamar for free, right? There's certainly some scoping issues there. So for example. It's a low end solution, it's for SMB. Yeah, but for the SMB market, this is a big deal. Yeah. You know, and you very rightly mentioned that for the virtualization specific backup vendors that are out there that have been selling to that low end of the market, some of their low hanging fruit just went away. Yeah, so I want to get into the study. So I want to pick up on something you said. You said DDoop is a must. Does your research talk to what percent of the base is using and adopting data deduplication? Yeah, so between those that are using and those that plan to use within 12 to 24 months, it's like 80 some odd percent. And I got to tell you, if you have me on the cube this time next year, that other 20%, that number's got to be lower. Those people just haven't done the math yet. So DDoop is absolutely a must across the board. When we did the research, so the project was actually called Data Protection Modernization. And so what we did was we looked at about five different vectors for how people are looking at changing their current data protection strategies and why. And that included things like DDoop, it included better use of the hardware, like for snapshotting and array based replication, included a big piece on how they're handling the virtualization scenario and then also what they were doing with cloud. And so on each of those vectors, we looked at what they were looking at, why, what they're doing now, why and why not, and then where they were planning on going next. So a lot of really good interesting stuff that came out of that. So what other findings did you have that stood out? So I think the one that probably stuck out to me the most was the non-stickiness of the status quo backup. You know, you and I've been doing IT for a long time now and for as long as I can remember, I've been doing backup for 22 years. And for as long as I can remember, backup was always like a religion, right? It was kind of like, you know, one of those things, once you signed, or political party, once you signed up for it, you were in for life with whatever that vendor was. Fossilized. Yeah, yeah. So the, about 50% of the respondent base for this last survey have had their primary backup solution regardless of version for three years or under. Really? Yeah, it's not like, you know, they've gone from V2 to V4 to V8 or whatever, three years or under. That's amazing. So virtualization had to be a big catalyst for that change, right? It was. It forced a lot of people to reconsider their solution. The other part of how non-sticky backup is, we actually asked folks, if you could go from scorched earth and start everything over again, only 32% of the respondent said they'd keep what they had. A third, right? It was like 18% said they'd go the cloud and just under 50% said that they would go to a new solution. Backup is not sticky. And I think virtualization as well as just, you know, unfettered, unstructured data growth, those are the big drivers around, they're going to have to reconsider how they're doing stuff and they're trying to do smarter. Virtualization specifically, the big challenge is around VM sprawl. You know, one of the things we saw in the keynotes we saw in a lot of demos this week is VMware and the other hypervisors for that matter have really made it just too darn easy to stand up VMs. You know, sub one minute at a time, new VMs get stood up and as soon as that happens, all of a sudden you're consuming gigs of storage. And oh, by the way, a lot of that is repetitive. The Windows OS in a thousand VMs still looks like the Windows OS. You just got to back it up. Right, so it's forcing DDoop. It's forcing a different change for how people are looking at backup. And so the status quo is not sticky, Dave. How did, Jason, how did cloud play in your study? So cloud is still certainly on the up and coming. So some basic math numbers on the two extremes. Tape is not dead, by the way. Tape is only in use in 56% of all environments. Cloud is in use in seven, okay. Now that includes, of course, some of the consumer angle that's out there, but and specifically when we looked at cloud, we looked at a couple things. We saw that direct to cloud, or what I call a D to C, was 2% of all environments. And then disk to disk to cloud, which I think is my preferred model, is at 5%, about two and a half times the adoption rate. And that makes sense, because if you look at environments of any size at all, they're going to want that fast recovery capability. Or more importantly, in fact, ESG is going to do a study later on this year. We're going to look at what I call Baz, DRAZ, and tertiary. Backup as a service, DR as a service, I think a lot of people are going to skip right over Baz and go straight to DRAS. And then cloud tertiary, meaning let's not throw the baby out with the bathwater, let's not get rid of that on-prem solution that we've made an investment in that's working well for us, but we still want to have the economics of cloud for that off-site copy. So let's use what we've got and take it to a cloud as a third instance of that data. So D to D to C comes in around 5% of the market. So you said earlier, tape is not dead, okay, you see that. There are some customers that have gotten rid of tape, but many, many have kept it as a last resort. Sure. You saw the Amazon Glacier announcement, I guess it was last week, which is, I guess, kind of aimed at those holdouts on tape. What do you make of Glacier? So I haven't gotten to put my hands on bits. I haven't gotten to see a demo on it yet. So I'll hold off on Glacier specifically, but here's what I want people to understand. It's never really been around tape is dead. That's just one of those exaggerated things that kind of kicks off a fun conversation. You're marketing from this guy. And this guy's liked that comment. But the point is, is that tape is also, let me say it this way, I think tape is dead as a preferred recovery medium. As the primary recovery medium is. It is not what you should be pulling data back from. It's a mistake to make tape a primary recovery medium, is what you're saying. That's your advice to customers there. I mean, if you're on tape, get off tape as a primary recovery medium. Yeah, if you want to use tape as a tertiary copy or cloud as a tertiary copy or tertiary disk, that's great, but really our data shows that around 78% of all environments disk is the primary means of recovery and then it goes someplace else as that tertiary and I think that's the right thing to do. Right, okay. And so let's talk about some futures here. You've had a chance to talk to some of the practitioners here, see some of the demos, listen to the keynotes. What do you see the future holding for VMware backup? So the thing that was most interesting to me is, so I've been spending a lot of time, once I saw what VDP was and I got excited about that, but I spent most of my time looking at private cloud. And when you look at private cloud, I think one of the things that's still kind of a gap when it comes to data protection as it relates to private cloud is there's a lot of ideas where basically the private cloud architecture says if we build it, someone else will back it up. And there's a couple different approaches to it. Some folks just use a brute force method of saying, oh, you launched VMs, I'll back those all up the same way, which I think is messy, right? Or you can just stand up VMs where there's agents inside and I think that's not scalable. Or, and this is the one that pet peeves me the most, is you go to this really elegant self-service portal, you bring up this private cloud thing, you launch this new service, and then you shut down that pane of glass and you open up something different and say now I'm going to go choose how to back it up. And there's just a gap there, right? One of the things that really impressed me about what I saw with VMware this week is with that sweet portfolio and seeing how they're really trying to look at the private cloud management stack, I think there's a lot more opportunity for them to say, look, this is gold, silver, bronze, as far as how something gets stood up. And that means, okay, so what does gold, silver, bronze imply from a data protection perspective? Maybe gold is protected every 10 minutes and retain it for seven years, bronze is protected once a day, retain it for 30 days. So I like that idea of letting the private cloud architecture decide the retention and the RPO, the RTO, the SLA. I want that to be an aspect of how private cloud launches. I think the thing that was missing from that was how does the private cloud architecture then go talk to that myriad of backup vendors and tell them, hey, I just got this new service, I just launched it, this is how you need to back it up. That glue is missing. Some of the things that we saw during the briefings this week around what the management stack looks like for vCloud director, what I think I'm seeing there is between that and what we see in VDP, there's some good potential where now that private cloud stack will know enough to at least tell VDP how to back it up and provide some constructs so the other guys can come on board as well. So that's encouraging to me. So you're basically putting forth a vision of data protection as a service. It's not one size fits all. Customer decides or maybe it's based on application or SLA decides what kind of protection I want. Do you also see snapshotting increasingly coming into the picture? Absolutely do you see snapshotting. But I'm going to twist one word that you used earlier. I don't think of data protection as a service. I want to see data protection as an attribute of a private cloud offering. So that whenever you define a service, quality of service of that whatever gets instantiated, the data protection aspects are just an aspect of that. And I think five years down the road, we'll start seeing how primary storage has that same effect and where you'll simply see RPO, RTO, recoverable points in time, availability, all just being aspects or attributes of the primary store. But until then, at least comes off with that. So you communicate with the private cloud service in those terms, let's call those business terms, and that sets the SLA for the backup. Exactly. Yeah, great. But let's talk about snapshotting for a second. You know, one of the things we saw in the data protection research was, one of the reasons I think that the status quo is not sticky, we saw that 86% of all backups are completing successfully, which isn't great. Better than half, which it was a few years ago, right? So it's not great, but you know, it's respectable for most environments. Here's one that bothers me though. 80% of recovery is complete within their RPO, RTO, SLA. So let's unpack that for a second. It's not 80% of recoveries, because that'd be a technology problem. What we measured was RPO, RTO, SLA. So after the IT guys sat down with the application stakeholder and said, okay, this is how fast it works and this is what my team can do. And they negotiate out the SLA on that. Even after that, they're only making the mark 80% of the time. That's a problem. And I think when you look at that kind of how can I recover it faster and where my limits are, I think the only way you get better than using traditional backup and recovery to get you better than 86% backup to get you better than 80% restores is to start being smarter about how you use the hardware in conjunction with traditional backup. And so I think that's where not only snapshotting comes into play, but also array-based replication as well. RTO, are they also missing RPO? They are. So against the entire SLA. That's really problematic. It is. One out of five times they're failing their SLA. It's unacceptable. It is unacceptable. And that's why I think the snapshotting is part of that. And in fact, we looked at how folks were doing that and about 55% of the responded base said they were planning on integrating either snapshotting, replication, or both in conjunction with their traditional backup and recovery mechanism. So this is a hardware plus software better together story. So that's like a great piece of research. How can people get access to it, get more information? So my blog's the best place. So technicaloptimist.com. Obviously all one word. That's my blog and you'll see all my write-ups from VMworld this week. I'll have a link to this video before the end of the week, hopefully. Oh yeah, absolutely. You'll be up on YouTube within hours and our man, Art Lindsay, he's the machine. Well, the report, the link from that is also up there and I'm gonna be doing some podcasting of my own over the next couple of weeks trying to draw out some of the more interesting graphs and sharing with the folks. Great, all right, Jason, hey, thanks very much. Really appreciate the insights. Thanks for having me. Always a pleasure. All right, thanks for watching everybody. This was the spotlight on VMware Backup Recovery Live from VMworld 2012 SiliconANGLE's continuous coverage. We'll be right back to wrap it up after this.