 Okay, we're back. This is Dave Vellante. We're live from EMC World 2011. SiliconANGLE's continuous coverage. You can watch on siliconangle.com as the blogs flow through. And of course, SiliconANGLE.tv where all the video and live feeds live with my co-host David Floyer of Wikibon. And we're with Mike Feinberg. Mike, thanks for coming on theCUBE. Thank you, Dave. So we're going to talk about Atmos. When I first heard about Atmos, it was like deep dark secret, right? You had the skunkworks in Cambridge. I don't know, David, if you were at that meeting. I was at that meeting, yes. It was in the Science Museum. Yeah, and we scoped out. We were taking notes like crazy. It was Hulk and Maui, and it was the greatest news thing. And I think you asked the hardest questions. I think it was. We asked a lot of them, that's for sure. And get that mic nice and close so everybody can hear you. So, you know, it was all about the excitement of a new way of doing storage. How's it been? What's the traction like? Yeah, it was a gleam in your eye back then, right? And we talked about, what interested me the most is, first of all, Tucci showed a little leg on the first day. So we got, we're in the cloud, baby. And you know, Tucci, when he says something, he's got a pretty good track record. And then, shortly thereafter that, you gave us a peek under the covers. And what was interesting was the design requirements, the billions of objects, the millions of users. And it really resonated with us. And then, of course, then you had to go to market and deliver and I'm sure you're learning a lot. What are you learning? What have you been finding? So I think, to your point, I think Joe was frankly instrumental in being the visionary of we need to do something different here. And he put the team together and I had the opportunity to lead the team to actually think about, how would you think about the billions of objects? And we used to call it the Carl Sagan storage. Storage with billions of billions of objects. But the fact was is that when you looked at what people were trying to do from an industry perspective and customer perspective, the technology we had didn't meet the need. So we actually had to look at the systems differently and when you took those notes, we really had a very good handle on what the architecture would be. Then we had to go execute on that. And the nice thing about Atmos is that we have introduced the product 18 months later into the marketplace, have been growing and shipping now for three or four years. And actually, thank you very much. Have been doing great traction with customers. And what you find, which is interesting, what you find is this challenge of managing unstructured content across multiple data centers and managing at scale is across a number of industries, number of verticals. And so what you might have thought was into the small part of the marketplace and certainly we're seeing traction with our service providers that are trying to provide storage as a service or building services on top of that. But also in the enterprises that really have to manage unstructured content at scale. So where's the sweet spot in the market that you are going after? I mean, there are a lot of people out there, a lot of different technologies. Pretty hard to do business with those service providers, isn't it? The service providers are a great partner. So of course they're easy to do business with. They don't mind, you know, you're getting a big giant premium, right? They're not good negotiators or any of those things, are they? Enough said, but no, here's the reality. If I think about what people are utilizing Atmos for, I think there's about two or three key use cases that I would highlight. And frankly, it is the service providers that are really focused on how I can deliver storage as a utility for their customers. And that might be within their data centers or it might be across the internet. And that is recognizing that you have to have a multi-tenant environment. It has to support multiple sites and be tolerant across failure domains and really be robust because frankly, when the system goes down on the internet, people might recognize that. And I think that's a first and foremost. The second part of that is service providers are recognizing as they deliver higher-level applications on top of these infrastructures, they need an Atmos-like infrastructure to develop it. So I would almost say that they're bifurcated on delivering a service based on Atmos and also utilizing it as part of their deployment of higher-level services for their customers. So that's one sort of leg of the stool, if you will, for the Atmos environment. The other one is what I would describe as the Web 2.0 or new application programming paradigm. And frankly, if you think about cloud, I would say EMC is doing a great job of addressing two of the major contingencies or components of cloud. One is how do you take existing applications and get them to work in a cloud environment? I think there are VBlock capabilities, it's doing fantastic there. But what Atmos is focused on is what is the new application paradigm? How do we write loosely-coupled application where storage can be anywhere? And why that loose coupling is incredibly important is if you're over the internet and you've made tight coupling assumptions, you're going to have a challenge. So the REST-based interface to storage, people are rewriting their applications because they have to have content across the world in multiple data centers and they have to do that in a way that scales from an administration standpoint as well as scales from a usability standpoint. So that's another leg of this stool. The third is we built a pretty good system to manage lots of capacity across lots of sites. It does that without a lot of management requirement. That fits the needs of archiving and we see a lot of people utilizing archiving based on cloud storage in particular with Atmos. And if you think about just this past week, we introduced having the Centera and XAM APIs on top of Atmos as an extension of being able to have that archiving play on top of Atmos. So it's sort of three high levels. So that's a great use case, Archive. Now, and of course the catalyst was all, I presume was all this web 2.0 buzz back in whenever it was 2006 and you're looking out and Tucci's pretty good at the gut feel game. And I think he saw that, but so that was a little bit of a surprise, I guess, sort of the way in which Atmos and cloud-based storage and object-based storage have been sort of permeating, right? I mean, it surprised you a little bit, but it's still, it's still a niche. I'm surprised that some of your competitors haven't jumped in as much. I mean, we feel, we've felt for a while like this is just going to explode. What's the big barrier there? So I think there's a couple of things I would say. And I don't know if I'd use the word niche, but I would say that it's homegrown in many of its usage. And so we got a couple of different things going on. If you think about the traditional storage industry, they're satisfying the enterprise customer. And the enterprise customer traditionally is focused on that transactional systems. And so we are seeing a transition from people just focusing on transactional to more content-related services and capabilities. And the whole notion of managing unstructured content is a scale problem. And so that takes some time to come to fruition. The enterprise is learning about that, and the leading edge of the people that are doing it have been these homegrown developers that are actually trying to change what, how they've done things in the past. And so, frankly, the industry hasn't focused on them. We see lots of startups. The main levels of the, you know, you do see competitive landscape. Some people trying to buy companies that are, I would argue, repackages of other technologies now describing themselves as cloud storage. But the fact is, is, do you design for cloud storage? It's really the startups, and I would say Atmos for all. That's for, frankly, Amazon, right, in Google. Yeah, so you got Amazon S3. You got some other sort of cloud guys. I mean, Nirvonix is out there. You got NetApp Purchase Bycast. We'll see what happens there. And some others, but Atmos is really EMC homegrown. I mean, you use some other technology as well, but it's in EMC, I call it a skunkworks. Is that a fairer thing to say? You know, I think skunkworks implies no one knows about it. When your CEO and chairman actually knows about it, I don't know if that's a skunkworks or a focused startup. That's his skunkworks. It's still skunkworks, you think? So I guess you're by your definition, skunkworks. Well, I guess you guys had facilities in Cambridge. You know, kidding aside, I think we were very thoughtful about how do you incubate and innovate a business inside EMC. Yeah, right. And I have to be honest with you. I think that's a great learnings and great lessons learned by that. But to your point, I think Joe is very thoughtful. Mark Lewis is very thoughtful, very smart about, well, we don't need to put this as part of the core. You need to separate that out in a way that's meaningful that actually allows you to innovate and incubate. I owned all parts of the business as we've gone and grown over the years. I don't own all parts of the business. I have to leverage the mainstream of EMC, which is a very good thing because I got a huge leverage point and consistency from my customer base. But it also shows how we've grown from the environment and how we are succeeding. Yeah, but to incubate, you don't want to make decisions that might be driven by maybe the need to keep an existing platform growing or you just want it sort of that freedom to make decisions independent of that. And I give, as much as I like to take credit, I have to give Joe and the executive management team and Mark in particular a huge amount of credit for setting me up that way to succeed. So this idea of spreading storage out, made some announcements there and geo-parody. And David, you were talking about how that really dramatically cuts the cost, right? Yes, I was very interested in that announcement today because that, your claim was very good and distributing slices of the data around the place and not having to do so much, like hold so much, something like half of it or 65% of it. Tell us more about that, is that going to be a very significant reduction in cost? Yeah, so I think a couple of things. I think it does a number of things at the end of the day. And just for clarity, what we announced today was an example, it was a production usage and people deploying this, it's not just PowerPoint, it's not just a thought process. All right, it's a product. We have real customers, real product deploying and we're basically taking one erasure code environment and erasure code as a way of protecting the data by segmenting the data into data segments and parody encodings. And at most we have some flexibility of 10 data segments and two parody encodings for 10 of 12, 10 data segments and six parody encodings for 10 of 16 and also a nine and 12. And the reason I use the nine and 12, which is nine data segments and three parody encodings is because we had a customer that took four sites, 50 kilometers apart and made one erasure coding data, copy of the data across those four sites. So 1.35 essentially copies the data or capacity usage and they got disaster recovery and tolerance to a site failure, any of these sites. And so when you think about a traditional DR, how many copies of data would you have to do for a site DR? You know what, I'm mad at my marketing guy, I'm going on record as mad at my marketing guy because at the end of the day, the way we did the math was we said, okay, minimally you would have to have one copy at each site, so four copies of the data and then you went down to 1.35 and it turns out to be that's the number we calculated. Practically speaking, you wouldn't have one copy, right? You'd have at least more than one copy at each site. That's right, yeah. I mean, they're one of your competitors, isn't it? It's ClevverSafers who's been doing some of this and they've been getting some good traction. Yeah, so we're seeing this to answer your broad question, it's greater than 65%. We think that it's applicable not only to the nine and 12 and the four sites, but six site configurations. There's a number of different configurations based on the encodings that makes sense. We had to do technology to guarantee that we have even distribution of the segmentation so there was data in all the locations and parity encoding so you could lose a site. So we did a lot of creative things to actually manage that fact of how people could utilize it. To your broad question, yes, we do believe that this is going to change the landscape of how people are going to think about disaster recovery. It's not just about active, active, it's active, active, active and reducing the number of copies and we think that in truth, there is so many use cases for this and you can take it to broader extremes of six sites and you can mix and match. You may have two parity encodings across 10 of 16s across six sites because you want it to have even more redundancy but it's still going to afford the cost. So the flexibility this is going to provide for customers is great. Not just about talking about it, not just saying it's theoretical in a PowerPoint, customers doing it and they can tell you about it, it's really very powerful. So you're using some proven math, mathematics to do this, right? And there's some other IP that you're layering on top of that. You guys got, have you filed for patents in this area? Can you talk about that? Well, certainly we are, I'll say at a broad stroke, we are always focused on making sure that EMC and its shareholders are protected from their IP. Yes, you are. So you can imagine that we're making sure that our IP is protected. Yeah, okay. What about security in this area? I mean, obviously you've distributed around six places or four places. Security must be quite an issue in this sort of environment, especially if you want to inter-mix stuff between the different sides. Yes, security is a... How are you tackling that? Security is a very interesting question in general in cloud storage. And I think that this is another example. So clearly customers can and may choose to actually not secure that environment or have the networking environment secure that environment. Or they may ask and many of the customers ask the outside people putting data into the system to encrypt the data. And when I mean secure the environment, I'm really talking about encrypting the data. Certainly securing the system. Is encryption part of your algorithm? Is that built in that? Right now, Atmos is relying on either external people to do the encryption and or some of the underlying systems we are in the future enhancing our system to have a little more encryption. In fact, in the GeoDrive that we announced has encryption into the system. So it tends not to be, believe it or not, a concern that necessitates that the storage technology does it. It has to be in the system. And many, many, many customers actually want to have their own capability of doing the encryption and not having as part of the underlying system. There's some obvious, I think you're well aware of laws in different countries that if you're managing encryption and you're doing the key management, people may not be happy with you when you get to serve some legal documents to serve up the data when you're going to have to apply. Let's talk a little bit more about enterprise cloud, object-based, rest interfaces. So the enterprises want cloud storage. We know that. There's some concerns about just putting all my data in S3. We've seen, we just saw recently, Amazon went down. We've certainly seen security hacks. That's not going to change. That's not good news for you guys either, is there a mechanism going down? I, you know, if you want to finish, I have a view of that. I actually think it is actually good news for these guys. I don't know why you'd say it wouldn't be. Why, what's your angle on that? Just because it brings down all of clouds. I guess, but at the same time, it says, look, we need a better alternative. And if I can trust a company like EM, pick your enterprise whale, EMC, IBM, HP, Oracle. I mean, if one of those companies is going to provide me a solution, I would think that I'm going to have more faith in that than, you know, Amazon is SLA, Joe Tucci said it best. It's like, you know, our SLA is we'll do our best, but if we don't, you know, don't call us. Or Amazon recently put in, I don't know how many people know this, a platinum level SLA, it's $180,000 a year. How many people could I hire, you know, or even outsource to India for $180,000 a year, a lot. So my angle would be it's actually a positive for you. I'm going to say it's a positive. I'm going to say in a slightly different way. I actually think that the reality is, and you're going to be surprised with this answer. So I think Amazon is fine. I think they did, they had an outage. You know what? Outages occur. And I think the point that I would have and why I think it's fine is it does actually say why EMC is different in its approach. And the approach that we're taking is one is a bad number. And biology, one's a bad number. And you know, data centers, it's a bad number. And service providers, it may be a bad number as well. You know, the end of the day, you got to make the assumption that S is going to happen and how are you going to react and design for that? In fact, I would argue that's what we did with Ask Most. We designed that stuff is commodity, it's going to break, what would you do? Take that whole design point of using cloud. So do I put data on my private environment so I can control it? Absolutely, that's some of the answer. Do I use a service provider and make sure that I have some of it internal and some of it external? Absolutely. Do I rely on one service provider or maybe I use a multitude of service providers? And why EMC and allowing you to buy on-premise, allowing you to buy from multitude of service providers, we're enabling that capability of essentially understanding your fault domains and how you can manage it. And just like in biology, one's a bad number and having enough to procreate is if I can stay with the analogy and I won't go too far, so don't worry about it. That's right, this is the cube, you can go crazy. You know, I think it's the right answer and I think the key though is in that environment, you don't want to have to have unique tools for every environment and by having Atmos as the underlying core technology and the API sets that we have, that's critical and that's what we'll be able to do. So I think it's a good thing. I don't fault Amazon, I'm sure some other service provider is going to have some other outage. It's a fact of life, let's deal with it and let's design around it. Yeah, but your explicit strategy with your service providers is to really, as I say, provide that enterprise class and you know, explicit SLAs that are more enterprise level. That's what your service provider is going to do. I would presume things like auditing, I mean, I can't go on and audit Amazon, I can't and we have clients in Wikibon, they have no choice, they have to do that. What about a security incident? You know, I mean, how do I define a security incident? I mean, so many things that your service provider partners have to deal with that Amazon doesn't. Amazon doesn't really care about that, right? And I think this is why we want to enable, and I think you said it well, the Amazon has done well for the developer community. The opportunity for all of us is the enterprise and how do you go after the opportunity? Well, the last thing you want to do is put somebody in a position where they have to make a step function or cross the chasm. And you know, we all know that yes, any enterprise got a customer and say, what percentage of your data can go into the cloud? You may get some number of, zero can go into cloud or 100% can go into cloud or 50, 50 or 70, 30. What's the right answer? All of them. And so given that construct, let the enterprise guy who knows his environment manage that volatility of how much they want to put in the cloud. And given the tools to do that, the way they want to do that and the way, maybe the auditing they don't like from the source provider, they're going to leave that on premise and they can do it. Maybe it's religious and it's not technical, but those are pragmatics. Yeah, absolutely. And we're trying to be as pragmatic as possible. And if I had a tagline, I would say, we're allowing IT to say yes to cloud. And they can say yes to the architecture and how they acquire that cloud. That's really their option and that's the flexibility. And so now, the other attribute of cloud, of course, is pay as you go. That's another thing that S3 sort of popularized back in the early part of this decade or last decade. Now, your go-to-market is largely with service providers. So they are in the business of pay as you go. That's how they're delivering this stuff. How about your model? You have a shared model where those guys are asking you to share in some of the risks, some of the pay as you go, and share in some of the reward or is it more the traditional model? Are we not there yet? Yeah, so here's how I would answer the question. I think this is an emerging market. No doubt about it. And then there's all creative ways to actually go after this business and making the right business model for the situation. The strength of EMC is the flexibility to work with its partners to give them the best model for their business. And whether they want to acquire the technology, they want to utilize it, we're able, or at least in financial considerations, we have a strong gamut. Here's the practical reality. If you think about this business from a service provider perspective, scale matters. And you want to get to scale. And once you get to scale, owning will probably be the asset, it will be the best way you're going to make the money. So it's like anything else. We have the flexibility and people have to pick how they want to go to market, how they want to do it when, and EMC's big enough to have all those options. So Mike, we have, go ahead. So you talked about the chasm, you talked about what you're doing in that area, but what's your breakout strategy? Where are you going from now? What's the things you're focusing on? How do you get to that critical mass of getting out? Yeah, so I mean, I think, here's how I would say it. If I've been shipping for three years, we see the right nice revenue ramp that every year that we want to see. Our goal is to make very successful customers, very happy, very referenceable, and being able to repeat. The absolute capability of the EMC, go to market, and the opportunity to have the reach is the goal and how to leverage that. And what I would tell you is if you think about it, that we don't disclose, if I was an external startup, I would be very proud of my first year revenue, second year revenue, third year revenue ramp. So I would be looking at, maybe not the Ferrari, but I would be certainly looking at the nice car as I felt Howard's succeeding. So I'm very optimistic about the growth. And frankly, if I could say it one simple way, we just need to execute. We make happy customers, the rest will take care of itself. So now Mike, we're getting your story, the vendor story that's always good to hear, because you're at the heart of this whole thing, but we got a customer coming on next. Now of course, they're coming into the cube, it's not rehearsed, it doesn't know what questions we're going to ask, so we're going to validate some of this stuff, but my last question for you is for the customers out there that want to go to the cloud, what's your advice to them? How should they get started? How should they think about this? So I mean, you would think that's a softball question, but I'm going to answer it a couple of different ways. One, there's those people that know why they want to go to the cloud, that's pretty straightforward, just jump in and get going. They're either doing it for new application development, paradigm, they want to be able to take advantage of the bursting capability. What I would caution is for people to understand what they're going to get out of the cloud. So my number one thing is understand why you're going to it, just because your neighbor's doing it, doesn't mean it's for you. Practically speaking, I think if you look at the use cases that we're solving and how the marketplace is evolving, and the introduction of device independence and iPads and playbooks and Android, all that, it's obvious that the cloud is going to be an architecture that people are going to have to enable. And it's having a very focused approach in how you can exploit it, so you can gain some early success and then have a repeatable success for your customer. So it's sort of obvious, I know, but you would be very surprised. I mean, maybe you wouldn't, of how many people say, let's jump into the cloud. What are you, and they don't even know what they're doing. I wouldn't, I think that's great advice because a lot of CEOs out there are saying, just get it to the cloud, just get it there, make it simple, and a lot of CIOs are responding, or maybe not CIOs, they're a little bit more cautious, but a lot of organizations are responding, hey, we have cloud, you know, check, but that's dangerous is what you're saying. So honest advice, I appreciate that. And I would tell you that EMC is very good at being able to explain the cloud and explain the sort of bifurcation, I talk about new apps and old apps, and we'll be happy to help you. So my last comment, by the way, it's always better to talk to a customer than the technology provider. Well, it's important to this dialogue. Both is good. Mike Feinberg, Senior Vice President, General Manager of EMC's Cloud Infrastructure Group. One of the fathers of Atmos. Thanks very much for coming on theCUBE. We appreciate it. Thank you very much. Good to talk to you.