 Okay, we're back. This is Dave Vellante. We're live at VMworld 2011. SiliconANGLE and Wikibon's continuous coverage of VMworld. It is an unbelievable event. 19,000 people here. Go to siliconangle.tv. Go to siliconangle.com. Get all the information on this event. Check out wikibon.org. We've got teams of people that are covering this show like a blanket. This is the storage optimization segment, and I'm here with Steve Kennison, my friend, colleague, and co-host sometimes. Hey, Dave. Good to see you again. Thanks for coming on. So, Steve, we're going to talk about compression, and specifically we're going to talk about real-time compression. We have these spotlights this year. They're in-depth segments. They're sponsored segments. As you know, IBM is sponsoring this one. You're with IBM. So, I want to pick your brain about compression, real-time compression, what it means, what you guys are doing, why it's different. I'm going to ask you some tough questions, but let's get into it. So, you were with StoreWise. StoreWise was acquired by IBM. Correct. Made an appliance, and you guys today are selling an appliance. That's correct. So, we have an appliance. That is what we call the IBM real-time compliance for NAS. It's an appliance that sits on the opposite side of your storage switch right in front of your storage array. So, as data gets passed through the appliance, we do compression in real-time. The interesting thing is, as we do the compression, we compress the data before it gets to the array, which, you know, so, you've got the right guy on. I'm not a sales guy, right? I'm a technologist, and I'll tell it like it is, right? So, while I do believe, and I will agree with you, and this is probably one of your tough questions coming up, is the embedded solution becomes maybe a little bit easier to manage, right? But, until they solve this notion of being able to do compression in real-time, there's not a whole lot of value to doing it on the array, because if you start doing it on the array, you start having a tremendous amount of system resource and performance impact. So, by us doing what I call compression offload in the wire, we can actually compress the data before it gets to the array, and there are a couple of benefits of being able to do that. One of those benefits is that, by shrinking the data before I actually move that disk actuator arm, I give a lot of those system resources back to the storage CPU so it can actually do more work in the same amount of time. The other thing is, by compressing it transparently before it gets to the array, whenever I get for a compression ratio, let's say just 2 to 1 for this example, I actually double my storage cache. And thirdly, because we actually do read caching in the appliance, we actually get read cache hits out of the appliance as well. So, we have customers with 65, 70% compression that actually get a performance boost because they're actually doing compression before it ever gets to the array. And that was one of my questions, is the whole embedded thing. Now, you made a statement, there's really no value if you're taking a big performance hit. Let's talk about that a little bit. There's value, but the value is not real-time value. It's more stale data, data that's not accessed. Is that fair? I think that's very fair. There are people who have implemented compression in a post-process manner on some of the competitive technologies like EMC or NetApp that are actually getting value. So maybe value wasn't the proper term. However, as the world, as we just looked in some of those charts before, slowly moves to a real-time world where I don't have this time to wait for information or access to my information and I have more and more users accessing my information, maybe in a big data environment or in a cloud environment. My question back to people who are doing it not in real-time is why aren't you? You better start getting on that bandwagon now because sooner or later you're going to end up playing catch-up and you don't want to end up playing catch-up. Well, a reason I ask that value question is because to me it speaks of an architectural issue maybe or a paradigm problem. In other words, why is that stuff on primary storage to begin with? It's a very good point. Why not through some kind of whatever, automated tiering or some kind of process, age-based classification criteria, whatever you want to use, get it off there and maybe compress it before you get it off. I mean, tape has had compression for years, right? Sure. So is that what most of this stuff does? Is it like tape-like compression or is it just keeping it on disk so you can sell more disks? Yeah, we talked in the spotlight a little bit before, right? So users are always afraid to throw stuff away, right? And then that's understandable. You never know when you might need something. The question is, is there starts to be a disparity between the information that you have online and available and your ability to access it and utilize it? And I think what we're seeing right now is people want that stuff online because of the what-if or just-in-case. The issue is that as I start to get applications that have the ability to access more capacity and do it faster, now all of a sudden I make that data what was not as valuable, more valuable. And the question, you know, we have to kind of wait for these analytic tools to kind of catch up to be able to have access to that information. But today, you know, IBM has five key technologies that they believe that you need to implement for storage efficiency. You have tiering, thin provisioning, virtualization, compression, and deduplication, right? I'd say all but deduplication today is ready for prime time for what I would consider real-time. I mean, if you think about the easy-tier technology, right, we have the ability to move information by just looking at a snapshot, not a snapshot, by looking at the data in real-time and moving it to the right platform. Virtualization and thin provisioning, I think everybody would agree, once you turn it on, it's kind of on. You virtualize and you thin provision, you know, you virtualize across your different data sets and you thin provision on the fly as you need to. Compression is really that next step into that real-time platform. So what's really unfortunate is the fact that we actually call it real-time compression. Here I am, the real-time compression of Angela saying it's unfortunate we call it real-time compression. What we've actually invented, right, is this platform that allows you to do almost anything real-time. So today we actually plug in compression. We use industry standard LZ compression. It's the same compression that Microsoft uses for its .x files. It's the same compression Oracle uses for, you know, the compression of its files. But what we've been able to do is the intellectual property on either side of that compression capability allows us to look at the information stored in such a manner that when we need to access it, maybe we don't need to go get all of the information. Maybe we just get pieces of the information and by only getting pieces of information you're creating a lot less I.O. We all know in VMware, right, I've got more servers accessing less physical infrastructure. If I can reduce the I.O. I can increase performance. So those things start to compound themselves and it ends up making the entire solution stack a little bit better. Yeah, so that's new information to me. This notion of you wish you hadn't called a compression is real-time blank. Right. So where else can you apply it? So, you know, through the acquisition process into the company, you know, the company had filed patents for being able to do things like quality of service, things like antivirus monitoring and that sort of thing. And there's a whole host of other things that we'd have the availability to be able to plug into this real-time platform. And I think the real key innovation here is this real-time platform. And, you know, as I talked about before, we know that users buy storage and buy their storage platforms for a couple of reasons, right? It's performance and availability, right? So whatever we want to introduce into those platforms, we need to make sure that we don't have an impact on performance or availability. So how do you do that, right? And you're going to see, you know, technologies such as compression and that sort of thing start to get deeper and deeper into the IBM product set and in each stack, we're going to make sure that there's no impact to those key characteristics for which users buy storage. It's unfortunate, I say, that we named it compression because I think, you know, competitors like EMC and NetApp have gone and set a market value of a technology called compression for free. And you don't really get anything for free, right? There's, as your core report points out, right? The closer I want to be able to get to be able to do this in real-time, I need to increase CPU, I need to increase memory, I need to increase, that's not really free, right? Our whole objective is, is how do you actually build a platform inside of our storage technologies that actually allows this stuff to add value to the user without having that tremendous impact? Well, I mean, I've talked, I'm going to test it. I've talked to customers from NetApp and EMC about their compression. They're very careful about where they apply it. In fact, if you read the documentation, they'll tell you, you know, don't just turn it on. You know, even though Dave Hitz will tell you, hey, turn it on, try it, but they'll be very specific that you got to be careful that it doesn't impact performance. I've talked to, you know, many of your customers, one in particular, a Burson engineer from ShopZilla. And, you know, that was enlightening to me in terms of the degree to which it actually improved performance. You know, the issue is, you got to get more customers. So, and I think that embedding that technology is really going to be a killer combination. So, I mean, are we talking, I don't know, maybe you can't say, but is it near-term, mid-term, or long-term that that actually is going to happen? Actually, in the very short term, you'll see some stuff come out around the V7000 technology that IBM announced last year, right? Last fall, right. Yeah, so the continuation of moving those five key, those five essential optimization technologies or storage efficiency technologies going into those platforms. There's other platforms that'll be right around the corner before the end of the year, and then there are other platforms that'll be longer term, right, where the real-time platform will be embedded. Excellent. All right, Steve, well, listen, we're out of time. I appreciate you coming on, and thanks very much for helping me out today with this spotlight. We've got a great panel coming up. This is the Storage Optimization Spotlight at VMworld 2011. I'm Dave Vellante, and I'm here with Steve Keniston. I'm with Wikibon.org, and this is SiliconANGLE's continuous coverage of VMworld 2011. We're here from Vegas. Go to siliconangle.com, siliconangle.tv, check out Services Angle, our newest publication, check out Cloud Angle, check out Wikibon.org. For all the coverage, we'll be covering this show like a blanket, SiliconANGLE, the worldwide leader in tech coverage, and we'll be right back.