 Okay, we're back. This is Dave Vellante and we're here at Wikibon headquarters in Marlboro, Massachusetts. I'm here with my colleagues John Furrier and David Floyer. This is the Cube and more specifically this is the Flash Cube. We're doing a Flash Cube on Flash. What is a Flash Cube? A Flash Cube is where we gather people and bring them into our studios. We bring in content that we've developed at other events. We bring in live guests and we comment on what we're hearing, what we're seeing in the market. Flash is one of the hottest trends in the industry and my colleagues John Furrier and David Floyer have been with me all day. Wikibon.org is the place where you go for research. Siliconangle.com is the reference point for tech innovation and you can go there for all the news. The Cube is where we bring you the best guests that we can find. We extract the signal from the noise. We go into events. We actually do stuff from our own studios and we are doing our first bi-coastal Cube live from both Palo Alto and Massachusetts. So we're going to give a summary of the day here and then we're going to watch more videos on Flash. So first of all, I want to bring in John Furrier to the discussion. John, you've been listening to our guests today. We had a number of folks come in live. Brian Bukowski. We just had Scott Dietzen on from Pure. We had Gary Orenstein on and a number of other folks that we've interviewed over the last several months. I want to ask you what do you see as the building blocks for modern IT infrastructure, John? Well, I think it's clear that the Flash ahead or the Flash ahead, this Flash Cube today, Dave, really highlights that everyone wants to move to this modern era or modern infrastructure and that's really driven by the cloud as a high level concept, which is essentially operationalize your IT in a very nimble, agile way and that is really driven by application workloads. That's the buzzword that's been kicked around, managing your workloads. So in all the conversations we're hearing that, for us, for average person, think about it as the mobile app as a way to get your mind around that. But under the hood is all the technology that enables that, the networking, moving packets around, storage, putting it on disk, or in this case, in memory, Flash or RAM, and then ultimately the compute power, the servers. So cloud is an instantiation of IT. That's driving it. So the modern era really is about a software led infrastructure where open source software can be deployed in a way that is most appropriate for the business, almost a customized set of infrastructure. So no more general purpose computing, no more general purpose. Here's a box and it works for 68% of the people. You can actually get customized solutions. So that's kind of, to me, the modern era. But underneath the cover, things that are really shaping the marketplace are databases, virtualization. These technologies are enabling a new way to do things, databases with Hadoop, unstructured data with structured data, all these things, low latency. And in the day, it's about performance. So the modern era is about software in a new way and Flash memory and Flash storage, it's the heart of that value proposition in that people who write the software now have new headroom, more capability, more headroom to do new things. So it unleashes creativity. That's going to shift to this modern era. So David, I want to talk about the horses in the track. We love the sports analogies here at Silicon Angle Wikibon. And we always like to talk about horses for courses. And you're seeing just some massive investments, huge valuations. We've seen a number of takeouts, in particular, recently, Texas Memory Systems and Extreme I.O. Neither of those was announced, but we're seeing valuations in the several hundred million dollars for companies that are just coming to market, three, four, 500 million, violin, reportedly, you know, doing an IPO at close to 800 million. And so, so David, what do you make of, of course, you know, Fusion I.O., even though it's taken a big hit, still, you know, a billion plus dollar valuation, it's been up over two billion in the past. So there's a lot of wealth creation here, a lot of high expectations. First of all, what in your view is driving that, and is it justified? What's driving it is quite simply the ability to do things that just couldn't be done before. It's being driven by the new applications being built on the new infrastructure, the Facebooks of the world, the Googles, the apples of the world, who are finding the constraints to what they need to do and having to go to a much faster flash infrastructure in order to succeed in the applications they wanted to deliver to their consumers. And following on that, it's the enterprise market that is going to be adopting those same standards as the price comes down, and we've heard that today in spades. So there's, and the longer term view of this is even brighter, because as new applications get designed in the enterprise to take advantage of this, there will be a, in my view, a major improvement in the ROI of IT and an increase in spend in IT, not necessarily in the area of flash, but certainly in the area of new applications and cloud service providers who will supply those applications. Alright, so let's talk about some of the new newer entrance. I mean, you've got startups, we just heard from Scott Deetson. We had David Cahill from Solid Fire on last week at the OpenStack Summit. These guys have to be their startups. You don't buy from a startup because you want to, you buy from a startup because the value proposition is so compelling, you almost have to buy from them. So are the startups better? Do they have enough of a lead? Now you see, you know, EMC buys extreme IO, that's a startup. Texas Memory Systems is not a startup. But it's got, you know, a pretty robust product. Do the startups have enough of a lead in your opinion over the IBMs, the EMCs, potentially Oracle will see what happens with HP. I want to talk about those in particular. But start with the startups. Who do you like? Who's interesting to you? You can even bring in the hybrid guys if you want the flash first guys as you call them. Talk about that a little bit. So that there are three major groups. And I think it's useful to take them on at a time. There's the hybrids as you discussed, the true hybrids. There's the flash only arrays. And then there's the PCI flashes and extension of memory. So those three marketplaces, I think are significantly different from each other. So taking the hybrids first, the key market here is in the SMB, the small end or the departmental system, if you like, where you want to have a combination of disk and flash. But where the flash itself is a high percentage of the total amount of storage. So if you take, for example, Tintry, which works in the VMware space, and this is will be will be going into the hyper V space as well. If you take them, they've designed a very, very efficient file system, coming into flash, allowing IO response times of less than one millisecond. And I'm providing a from a cost perspective, a very, very cost effective type of service. Now they've been very successful in their small area of VMware, very successful in bringing out products. They've, to me, our prime target to be to be purchased because the, the hybrids that EMC have and IBM have, and, and NetApp have, they aren't really suitable for this new environment for the fall. They need a true hybrid if they're going to go after that marketplace properly. So agile is another member of that. Who was that? That was as you say, Ted, or it is the leader in this space. Sorry, can you come? Sorry, David, you cut out there. So you said, Tintry, do you mentioned, was it Tejial? Tejial, yeah, I think I said agile. I mean, Tejial, and Nimble. Nimble is clearly the, is the largest leader in this space. And I think at least one or two of those will be purchased by the whales. It would be my prediction. And come in as an SMB type product for those companies. And your premise there, so there's the, the flash hybrids you're saying are suitable for the, the small and mid-sized business. Is that what you're saying is for the cost sensitive folks? Or for a particular department, or for a particular small, small part of a large data center, great products, but they, their scalability in the same way as the flash only storage vendors have, they have less scalability. And they're already a combination of flash and storage. So they fit a particular niche in that SMB very well indeed, in my, in my opinion. So I have a question, David, on Twitter. If you're an SMB, what are the criteria to think about knowing the expense of flash when deciding if an all flash environment is correct for you? What are your thoughts there? I, I, I, for SMBs, where you get need some degree of disk anyway, having a single product that will solve the disk and flash, it makes a lot of sense. And, and go for the, the, an architecture which allows you a lot of flash, the percentage of flash you should be aiming for is around 20%. Okay, so now let's, let's talk about specifically some of the new entrants into the market. In particular, I'm interested in, in, in extreme IO with EMC and the Texas memory systems. Let's start with extreme IO. There's, they're currently in very limited availability, they call it directed availability. What's your take on, on extreme IO? I mean, EMC has a way of just coming in and dominating markets. Do you expect the same with what they're doing with extreme IO? I, extreme IO is an architecture is a good architecture. They need to add on to it. The tier one capabilities, they're, they're very limited at the moment. They just have some basic capabilities. But EMC have got a tremendous track record of taking products, making them tier one, making subdivisions of them for other, other marketplaces, and being very, very successful at rolling those out, making them high quality, giving them a worldwide marketplace, giving them the volume. So I would expect among the, among the winners in the flash only storage arrays, that EMC's extreme IO would be one of those winners. The other one is I want to talk about is IBM, the acquisition of Texas memory systems of 30 year old company, IBM buys the company, they blue wash the company, put it into their system, and then, you know, catapulted IBM last quarter, indicated that it was struggling in the storage business, but it said it was looking to flash to really buoy up its business in the second half of the year. What about IBM? I mean, they've got the distribution channel, they've got the, you know, IBM Global Services presence. What are your thoughts on the degree that they're going to succeed in this space? Well, I think IBM as a services led company is again the best as as a services led company. It has very good relationships with senior executives. It understands how to put it in business terms. And I think they will be able to paint a picture, an investment strategy for long term for their customers very successfully. People will want to start at a lower end, smaller end of the marketplace, you know, just getting their current applications working well. They trust IBM as a supplier. They have a worldwide spread. So again, I would expect IBM to be a significant player. In this space, they do need to do they need to bring in their MLC, they need to bring in a tier one capability other than putting it behind the SVC, the, the, the product that use at the moment, they've got to move hard and fast to get that to a state where it competes with the other off-mash vendors on price. You're saying, you're saying they need a, they need a richer data management stack storage management stack bringing in maybe their real time compression capabilities, and bringing that into the Texas Memory Systems product. Is that a, is that a bolt on? Can they elegantly do that? What are your thoughts on that? I believe that they will be able to do that. The, the fundamental architecture of the, of the Texas Memory is, is thought is very sound. And I think they'll be able to put that on. They're already have MLC products in the marketplace. I think they'll be able to, to, to pull that off. But it's the time scale. They really are going to be under the gun to get that stuff out fast because EMC is going to be moving fast already. There are three, four, five players out there in that flash only marketplace. And if they don't move fast and really get, get hold of it at the nascent stage, they're going to be in total catch up mode for many years to come. Well, and I think that the, the violence, timing of the violin IPO is critical. I mean, that, to me, violence got to get to IPO. It will, you know, give it the presence and another injection of cash that it will need to compete. It has to spend a lot of money to compete the way it competes, which is a lot of belly to belly. Obviously they got relationships in the distribution channel. But, you know, again, I said earlier in this broadcast, the violin seems to know how to sell, but they really pour money into that channel, that customer interaction, that service capability, they're building out, you know, their product capabilities. What's your take on, on violin? Yes, violin is, is, it sells on speed. And it's a sort of halfway house between the traditional all flash array, and it's got many of the characteristics of an all flash array. And the higher speed flash is an extension of memory from people like Fusion IO. So violin is, is certainly sell speed. It sells very effectively in that area. The long, the longer they can make a very successful play in that space, if they get sufficient volume, what their threat is that they get attacked from Fusion IO and products which use the Fusion IO type architecture, which EMC and, and IBM are working on. But they have a good two year staff. So if they can bring down the price of their product, bring down, bring down the, bring down in terms of latency, improve the performance of the product, they can be very effective in that marketplace. Okay, whether that's the, whether that's a 10 year play, I'm not sure they may need to, to, to rejig their architecture at some stage. Okay, we're, we're a little tight on time, but I want to, I want to talk about HP. So I had a conversation, both public and private with Meg Whitman, several weeks ago. And I asked her, you know, about uses of cash, stock buybacks, acquisitions, she said, I'm going to buy anything until we pay down that debt. I think the debt's, you know, over $20 billion. I talked to him private and said, Meg, are you really going to be that dogmatic about acquisitions? I mean, it just seems there are so many opportunities out there to bring in inorganic acquisitions, lower the risk of R&D. And, you know, there's been some, some positive signs out of HP labs, but you know, we're still looking for that huge home run in the last, you know, several years. What's, it seems as though HP is going to be relying on three par for an all flash architecture, as well as, you know, left hand all flash array. What's your take on HP? Do they need to make an acquisition? Can they actually adopt the three par architecture and compete in an all flash from a cost standpoint? What's your take on that? Well, time is going to tell. And I think they've taken the most difficult route of all in getting there. They, they can make an all flash module. They can redesign the, the stack itself to fit into an all flash environment. Almost every, no, not almost all of the flash only startups have started with a clean sheet of paper, because they needed to do things in a very, very different way. That can be done by HP. They really are under the gun to again, get into the marketplace with something that's going to work and be, be effective. I think for a long time they thought that memristors would replace flash. I think that's looking less, less likely. I mean, maybe they can pull that off, but that again would need a major investment of money in that space. I think they still feel that way, David. Sorry to interrupt. I think they still feel that way. Martin Fink at the HP analyst day was holding up memristor and, you know, touting its potential. I mean, I think, still think they hold a lot of hope out for memristor. You're skeptical and I think you're skeptical mainly because you're a believer in the consumer flash volumes and the economics, right? It's purely about economics. In my view, unless there is a consumer marketplace for memristor, it's going to struggle and struggle and struggle to, to go down the learning curve and, and be a cost effective. If you look at MRAAM, which is a technology that's been out there for two or three years in that same space, it just hasn't been able to get the volume it's required in the enterprise space or in the space. So, and I, it's, it's a complete change of, of idea and redesign of a whole computer. If you take the memristor approach, HP might be able to do it, but it would take billions of dollars of investment. And if they could, if they had those billions of dollars, I think they could spend it better in some other ways with flash. What about, what's about NetApp? What's your take on NetApp real quick? Well, NetApp have got a great architecture, a great scale out architecture. The way that they've tackled the marketplace is through caches. And caches work very well for read only and for small amounts of cash, but they don't scale to the 20, 30, 40%. That's, that's not where they're going to be. The new architecture of the, of the boxes coming out, looks good, looks very good, flash only architecture. Again, they've got to get that out and they've got to make sure they can integrate that in with the, the great sets of data, of storage services that they have with on tap, etc. They have on their current system. So their challenge is getting the product out and adding on to it all of the features. Again, they're late, like with HP, they're late, they're two, three years late. So again, there's an if you look at the opportunity for the startups, it is to go into that white space around the HP accounts and around the NetApp accounts and, and, and provide an alternative pathway quicker to a high percentage of those customers. That's the risk that NetApp and HP have over the next two, three years. Now what about HDS? We don't talk about HDS that much in this space. They don't make a ton of noise. They like to fill sims, you know, a former quarterback of the New York Giants, right? Reliable, they, they win games, they show up, they don't fumble. You've made comments in private to me that you like the HDS architecture. They have this sidecar, I'll call it. Maybe you can explain it better than that. But talk about HDS, where they fit in this whole thing. Well, they, they, they start off by having an SSD technology. They put a lot of effort into the SSD and done some very nice things within it. So they start off with their own SSD itself. They've been able to put around that, a controller, and, and a stand-alone flash-only component of their, of their product. So, again, that's early days. It's not out there in volume. If that, if they can show that that works, and it's going to be a similar way that HP is going to have to do it with 3-part, if they can show that that works, the big benefit of that approach is that they have already the software stack, tier one software stack that they can use and accelerate that time to marketplace. That's a big if, but if they can, then they do have a way, and they have very happy customers who are very happy with their software stack. So, what, what I believe they're doing is, first of all, maintaining their marketplace to their existing customers, they put that as a priority. And that's a sensible strategy. Okay, I want to get a couple of rapid-fire questions for you. Then I want to go back to John Furrier, give him the last word for the wrap. But, David, what happens to traditional tier one storage, VMAX, DS8000, Hitachi VSP, 3-par, you know, traditional storage. You know, they're a little different. They started with the sort of, they started the whole tier 1.5 trend, but have eaten into the traditional tier one. What happens to tier one as a result of all flash arrays? Well, the price differential they can charge for the software goes to hell in a handbasket, basically. They're good products. There will be a need for storage tiers at a lower level than flash, the passive data. There will be a need for that. They will, they will, a lot of legacy applications that will need the storage continue to use those storage for many years to come. This is not something that's going to suddenly change, you know, in one year or two years. So they will be able to sell those, but the price differential for the software will go down rapidly, very rapidly, after a couple of years, or one year or two years. And they will have to sell really pretty well giveaway of the software. So, now, that's the way they're going to have to continue. Last question I have for you, and then I'll turn it to John, is Fusion I.O. You've been a fan of Fusion I.O. since 2009, since you first saw that architecture, the server-side flash. We talked a lot about function moving back, you know, to the host. The best I.O. is no I.O., all the cliches that I love to throw out there. But now that, you know, they're under a lot of competition, you've got EMC coming after them, it looks like they're going to be doing atomic rights. You know, you've got rumblings that IBM's going to be doing that. You know, granted, NVME is off, a ways off. It's kind of stuck in committees right now, but we're talking about Intel. What's your long-term prognosis for Fusion I.O. dominance? Or are they going to get crushed by the big guys? It's going to be one or the other. If they can take their technology to the marketplace and get adoption, not only in the Linux space, but in the Oracle and the DB2 spaces, in the database spaces, they could become the de facto standard. The quality of their software is far better than anything else in the marketplace in the specific area of very, very low latency. There will be applications that can be designed in the hyperscale area that they and they alone can actually contribute technology to. So if they can get that adopted as an open source standard and be the de facto standard in the hyperscale marketplace, they have a chance to really make it big. Will EMC and IBM catch up with them? Well, if you follow sports and horses and cycling, etc., there's a pretty good peloton coming behind Fusion I.O. It's out there, but it's out there alone. So there's a significant chance that it'll be overtaken. But there's no doubt that their technology is two, three years ahead of anything else in the marketplace. So David, first of all, I want to congratulate you on an outstanding job that you have done in the space. When you first saw Flash come out, the EMC haymaker, you got all over this. You said this is a huge trend. It's going to change the industry. It's going to change system application and database design. And it appears that you were correct on that call. I think you've done an absolutely fantastic job of researching this market. So thank you for that effort. I really appreciate it. You're welcome. The other thing is a lot of questions. We get a lot of questions about how much of the data is going to be on server side Flash versus all Flash arrays versus hybrid slash traditional arrays and our forecast shows that a small piece of the data will actually go on the Fusion I.O. like server side Flash, maybe a single digit number in terms of terabytes, but around 20% of the spend is a lot of high value stuff. And around 45% will be in that so-called bit bucket and the rest will fall into that sort of hybrid, oh, sorry, all Flash space in hybrids, Flash first, all Flash arrays. So interesting mix. You can find this research on Wikibon, wikibon.org, free research, peers helping each other out. Go to siliconangle.com, the reference point for tech innovation. This is theCUBE. John Ferrier, if we haven't scared you away with all this Flash inside baseball, I wanted to give you the last word on the wrap. Well, I liked the conversation. I think it's spot on. I think it's some of the best analysis in the business, no doubt. And it's free too. I find it ironic that the best analysis in the business, the most disruptive area is free. And I thank Wikibon. Dave, you do great work. So it's just so much fun to be part of open source content. And again, we are all free content, SiliconANGLE and Wikibon. So great job, Dave, and your team. My big walk away from this day is it reaffirms what we already have been talking about, but puts an exclamation point upon the cloud. Amazon Web Services and the cloud and the enterprise are all colliding and it's causing a lot of massive movement from the big whales. And so the key walk away there is that's happening. The modern changeover is moving over. What I learned today, though, was a supply chain comment David Floyd mentioned in passing, but really to me that highlights the differentiation for big data and the infrastructure. So the supply chain pricing, supply of good quality media, flash media will be critical. And that's going to be a big element and a big variable on how these companies make management decisions how they organize their business models and ultimately if they're going to be profitable or not, if they can move fast enough. So, you know, having the product is a key component. The software models are actually continuing to talk about the technology advantage as David was mentioning. That's a big thing. And that to me, I think the lead is shortening for the for the leaders and the big guys are moving in and using that supply chain. That's an area that we're going to watch closely on the silicon angle and obviously the business benefits of economics, functionality, speed and performance and the value of value creation. So again, we'll follow this on silicon angle day to day. A lot of great original content there. Dave, and I think that's the key walk away here as we heard it from the executives and the leaders in the industry here today on the FlashCube, FlashAhead. So it was a really great day. John Furrier, David Fleur, thanks very much. And Kenny, really great job today from the remote studio in Palo Alto. Really the first major broadcast we've had from that studio and the first simulcast bicoastal broadcast that we've had. So the folks in Palo Alto are going to sign off now. Again, thanks guys. We're going to keep it right here. We're going to go to more videos that we've captured from the 411 IBM event. Right now we've got Brad McCready up and Brad is the vice president in an IBM fellow and someone who's taking that full systems view. One of the things that IBM has, in my view anyway, struggled with in the last five, 10, 15 years is bringing that storage and server piece together. When IBM made the organizational change under Steve Mills to bring systems and software and storage together, that to me was a milestone. And I asked Brad to the extent that they can actually begin leveraging that. Let's listen what Brad McCready has to say.