 Okay, we're back. This is Dave Vellante. I'm here with John Furrier and David Fleuer who on the West Coast and our Silicon Angle Studios in Palo Alto and gentlemen, we just heard from an IT practitioner I mean essentially David as you pointed out these guys are just trying to keep the lights on they're trying to make their existing applications run Faster databases the big challenge And it's interesting to note that Karim Abdullah of Sprint is putting TMS the what's now called IBM flash systems behind a sand volume controller now part of the reason for that is David you've explained to me that TMS doesn't have a rich set of software yet IBM hasn't ported anything over the data management volume management storage management stack So they're essentially bow guarding the stack from sand volume controller, which is pretty good Talk about that a little bit. Is it a viable long-term strategy in your view? Oh? As a short-term strategy, it's fine And it has the advantage If the if as I suspect to Sprint has a lot of SVC the storage volume controller that they're probably using that as the standard set of tools both for replication for For thin provisioning and all the other added value storage services At the moment, this is true across the board. There isn't really a tier one There isn't a tier one flash only device out in the marketplace So if you want those facilities, you either put them behind an SVC. That's not ideal You'd like to be able to do that directly You'd like to be able to map those type of functions onto how the flash could most be efficiently used Those are coming EMC is obviously investing them IBM is investing them HP Everybody's net up all investing in those types of capabilities But they're not here yet and the SVC is novice way to go in the short term So David, what do you see for all flash array pricing? Where are we at? Don't sort of dollar per whatever gigabyte? well, it depends on the type of on the functionality and on the type of Technology used there's two major types of technology SLC flash and mlc flash The mlc flash is the same as consumer flash and half the price or even less SLC flash is much more the the the enterprise level flash So if you're if you're looking at SLC you're talking around the $15 per gigabyte that sort of order you're talking about using Critication and compression and mlc you're coming down into the $5 a gigabyte or lower Around $5 a gigabyte that that type of pricing if you are looking at the The hyperscale market, which is a very specific marketplace where you're designing things the last only a certain number of Years for example, you're getting down to the $390 per gigabyte in that sort of area So it it's Dependent, but that's the sort of range that you're getting at obviously disk is much much lower than that you can get this sub one sub gigabyte Subdollars per gigabyte So that's clearly volume disk is much lower, but from a performance point of view flash is really at the same level as high-performance disk as 15k disk the the SAS 15k disks are around the same price, okay? So and I think I'm correct in that I don't think that the IBM flash systems what used to be called Texas memory systems I don't think they use mlc. So of course IBM is focusing on reliability They're focusing on focusing on cost per IOP, which is smart Because they're they're probably not the lowest cost out there And I don't believe they use compression and D-dupe yet Which eventually is going to happen they bought store-wise a couple years ago that's going to move in but John Furrier I want to go to you and and see what you're Hearing in the valley in terms of pricing in terms of supply You know there aren't a lot of suppliers of NAND flash What are you hearing in Silicon Valley? Well Dave, I think the supply chain as I mentioned earlier is a big competitive advantage So you have a lot of people here who are of interest in that On the data center side you got Google and Facebook they build their own stuff Especially Google and on the supplier side you look at Intel, okay? You have violin memory systems here in Silicon Valley, although Fusion IO will hear for them later They're in Salt Lake City an hour away on the plane, but huge issues here and again this comes back down to managing the strategy of Architecture and component buying what I'm hearing in Silicon Valley is that Intel's having some struggles On on the quality and the amount of NAND they could get as well as Violin has locked in it's been reported they have a relationship and essentially cornering the market now that leaves a lot of Opportunities some are speculating that a new player is going to come in at the three dollars price point underneath that and just decimate the market We have a source out there that we're trying to get some reports from We might actually have some commentary and confirm that later around who that player might be we don't know We're investigating that here But what you're hearing here is is that all this flash discussions all fine and dandy we could talk about it all day long But there might be a serious shortage in the marketplace and that's going to increase the prices and change some of the dynamics amongst the big players Yeah, now micron is obviously a big player, you know Boise and and you mentioned violin They've got a partnership with Toshiba all the many of the flash guys are trying to lock in well I don't want to say I don't want to report you But I think micron might be the one we're trying to get a confirmation on that but that's that's what we're hearing Again Dave, you know Supplies an issue you can use a ton of supply But if they if the quality isn't there you're gonna have a lot of Product failures in the field and we've seen this movie before I mean the bubble go back to the 80s the PC revolution Remember the Winchester disk drive for all the young ones out there. They don't remember those days It was a brutal bubble and so again We're seeing same kind of dynamics of that that Winchester disk drive back in the PC revolution There will be might be a flood of quality problems on the flash media itself Big red flag and all the vendors are paying attention. Yeah now the other piece of the action here the emcee is Announced extreme. I owe they announced cards, which are reportedly the Vera dent cards We asked the emcee to come on to this program, but they they declined just I guess too much stuff going on You know just wasn't the right timing for them Of course, it was short notice, but but we we did invite emcee on but but they are selling why why I asked question Why didn't they come on was it they said they were busy? I mean come on I mean they're doing a lot of controlled messaging So, you know, I love emcee's launches They do a flare and they try to control the narrative as Jeremy Burton always talks about but you know This is an important area. What what was the real reason why they stayed at home? I don't know. I mean, you know, it was short notice admittedly We decided to do this flash cube on short notice. I reached out to a number of folks as we know I got many people who wanted to come on some couldn't Emcee you know guys basically it was just I think too short notice just got back from vacation last week in the east coast And we just couldn't pull it off But what I wanted to say about those guys and I don't know you maybe they do we weren't comfortable You know that they couldn't control the narrative. I'm not sure but you'd have to ask them that but but nonetheless We'll have them on I guarantee we'll get them on we're gonna be at emcee world and I'm sure we get them to talk But supposedly they're they're OEM in the virulent card the point I want to make there is you've got Seagate and viridant As in the supply chain selling to emcee now David Fleuer We know that emcee is not gonna lose on price They are gonna get footprint and they're gonna sell software and they're gonna sell services So that is a disruptive force. You've got violin who we're gonna hear from next We're gonna hear from Don Basile their CEO trying to lock in a supply chain deal with Toshiba You've got other guys, you know fusion. I own others locking in supply chain deals The pricing is gonna get really interesting as John said supply could be lower You know, there's maybe a little cartel amongst the makers here And then of course you got Intel looming, but they're not moving fast enough So I don't know if you guys have any comments on that if you do great And then let's go to to Don Basile in a moment, but David any comments on that Sure Every technology goes through these little bubbles and back down again If you remember a couple of years ago, we had exactly the same issue in terms of Interruptions to the to the flash marketplace and the the flash goes up for a time and then comes down as new technology new Lower nanometer Technologies come in The the the important thing in from the seagate perspective or from the the storage perspective is that all innovation in terms of New levels of high-performance disc have all dried up the all of the attention in the in the disc marketplace is to make Inexpensive tubs of data as much data as they can get into it so that really That that's not going to be a solution to go out and go for a higher-speed disc That's that's not a solution long-term in the marketplace. So then we'll go up and down, but the overall Thrust is that nan prices will come out From a performance point of view to be a lower price than the equivalent discs And they'll take a larger and larger percentage of the high-performance market the active data market the passive data That will all be on this. Yeah, okay. So so we're seeing a big range of pricing from $3 per gigabyte in the in the PCIe Hyperscale market all the way up to as high as 15 for the non mlc non-dduped Non-compressed big range. One of the players here is Don Basile We've got a video of Don from from last year. He stopped by at EMC world and talked to me You'll see I botched his name. I called him Basile like my friend Peter Basile But Don was very gracious and and didn't embarrass me But but anyway, let's watch this video from Don Basile of violin a player That's Reading to do an IPO and somebody who's clearly figured out how to sell into the marketplace Let's watch this video. Hi everybody. This is Dave Vellante. We're back. We're live from EMC world 2012 And this is Silicon angle dot TV's continuous coverage of EMC world as I said at the top of the show We've got a simulcast going on at the H base conference in San Francisco My co-host John Furrier flew out to that last night, so we'll be Simulcasting that go to silicon angle dot TV check the H base link if you want to watch that program, but we're here live We're here with Don Basile the CEO of violin memory a violin memories one of the hottest flash startups on the planet right now Don has been growing the company Has a strong background a former CEO and chairman of a fusion I owe has raised I believe 190 million we're going to get into that for violin memory getting ready to do an IPO Don Dr Basile welcome to the Cube. Thanks for having me. Good to see you. It's great seeing you. Thanks so much some time here We're here as well. Yeah, we were talking off camera. You said this is your first EMC world It is it is I like VM world I go to that many times But my first time at EMC world and I hear it's a record for EMC. It's quite an event Isn't it now I can ask you are you here as you know like a v-plex sort of partner? Are you here as a friend a foe? Well, one of our partners the ICI Has us in one of the booths here. Okay showing off a great VDI solution using violin I see other guys from Marlboro, Mass, right? Yeah from Marlboro, Mass. Yeah, long-time big EMC distributor and Need some flash to go ahead and make those VDIs run and they're Demonstrating violin in the booth. So really excited to come down and support them. Yeah, good I mean flash obviously is hot Joe Tucci's been talking about it talked about on his keynote We saw the acquisition of extreme. I owe that had to make you happy What a great valuation was it for 20 million for 25 people to remember? I mean, that's we have almost 400 people. So when you start doing that math, we're we're getting pretty excited You start to get the linear equation out. It's good. It's good But yeah, so I mean a lot lot going on in your part of the world before you saw the Facebook IPO, obviously We did really what you think about that. I think it's hundred billion dollar IPO in the Valley Six-year eight-year-old company. It just shows that the innovation is still possible I mean people coming out 2002 were like is the internet dead is it all been done and then a Facebook comes along and creates a whole new ecosystem. So very very exciting area and you know, it's an area Violins not directly in that particular company But it's an area where I have personally invested in the whole area of social commerce Which I think it's going to be a huge explosion, you know coming on that ecosystem that Facebook has Google Microsoft have around the social connection. So exciting time really to be in tech innovation again. Yeah I mean, it's it's really the third mega IP, right? You obviously Netscape and Google and now Facebook and although Facebook's taking a little heat It's first full day of trading it pulled back I guess Morgan cut its estimates, you know while the road show was going on So a little controversy there but you know we would expect that data as the really the central point a central leverage point of of Facebook even more So than potentially advertising. It's all about data and And and you know, we think there's a lot of potential there We have we we agree with you really exciting time in tech. So let's talk about violin You guys have you know, it's reported to be an 800 million dollar valuation based on the math that we just did with the Back of the napkin and stream IO. It's a little undervalued right now, but Still pretty good. Well, I think you know, that's an equity investment value. It's not an acquisition or IPO value So, you know, I think that I think the equity investment value of extreme IO has played a bit lower when you're bought, but I think EMC has been one of the big buyers in industry They've paid some some good dollars for Islam for data domain Now they paid a very high price for extreme IO at least for all the people in the company So I think it's exciting, but I think what it's indicative of is that The value to the enterprises in the end of the day Violent exists to make applications run faster and cheaper for the end user customers That's just that's what we do. That's what we think about all day long And it shows that this new approach of memory-based technologies is what the end customers want And it's a legacy approach is you know in in danger of being replaced And so the companies have to either build themselves is very hard to do the violin IP goes back to 2005 Another stream we got from 2006 or talk about six seven years of development to bring it to fruition or that to buy their way into the market and So I think you're seeing now, you know, this is just on the heels of a number of acquisitions supply and acquisition the sand force acquisition Extreme IO acquisition people trying to buy their way into the market so they can start to make products And I read the the newspaper report messed it in like next year. So sometime next year They hope to turn that technology acquisition to a product acquisition. Yeah, they're all fleshier. I mean it's it's curious I mean we're watching thinking okay Is there going to be you know in the NFL when somebody drafts a left tackle all of a sudden there's a run on left tackles You know, you think it'll be a similar sort of run on on flash companies or is it a little bit too early for that? You know, I like in this space First of all space is so exciting. I mean, I haven't been this excited since the networking space in the 90s And I like it a lot to that industry and that industry you had many many many companies formed throughout the 90s There was a big incumbent called IBM and IBM and technology called s&a and almost every company used s&a 70 80% of the data flew through through s&a and in the corporate network even the middle of decade was still there, but out of that decade 20 30 40 companies were bought for hundreds of millions or billions of dollars and the end Cisco emerged and IBM actually sold S&A to Cisco and sort of exited the the networking business. They just got back into it another 10 years later Yeah, so I think that it's very interesting here. We're seeing a similar thing There is another 30 to 50 companies already formed venture back that are trying to do it violence there Okay, and create all memory arrays all flash-based memory arrays to go ahead and bring it to the market much like there are 30 50 100 companies creating IP networking devices in the in that so I think that it's it's healthy It's very healthy and it shows that in the end the customers of want the technology And they're looking for the approaches to get the technology list from an incumbent vendor or an upstart That's to be determined What we can do is kind of try to copy with what the great John Chambers has done Which is basically build violin make it bigger and bigger and bigger and keep servicing the customer over the next Five six seven years so that we hopefully emerge as a very important player out of that out of that period Well, I I wonder if I could run this by I kind of look at violin as the data domain of flash In other words data domains value proposition was you know, obviously it was the head It's data duplication piece, but a real appeal of data domain was he could drop it in and you have to change Anything you have to change your processes. You guys talk a lot about no changes to applications So I use that analogy because it's disruptive, but it's not it's disruptive the industry But it's not disruptive to the it practitioner. Is that a fair angle? It's a brilliant insight and it's something we work really hard about, you know, we we talked about 80 CIOs When we first took over violin is violin 2.0. We took it over recapped it And they said listen, don't make me change my infrastructure Someone has always given me something faster and better. Just let it plug in. It's why we have the flex certification SVC certification work with Oracle ASM semantic veritas You can plug violin today get the benefit don't have to change your database and make it all happen And that's I think the reason we've been able to get some of the biggest customers in the world To ready adopted violin such a short time period now you guys are readying to do an IPO right talk about that a little bit Why IPO I think the public market has shown and you know hats off to the fusion management team today And that they're basically able to drive a multi-billion dollar valuation in the public market able to access capital I think that that we're being asked to do the same thing our customers our partners are saying go public Raise another few hundred million dollars of capital grow violin even faster to go ahead to take advantage of the opportunity Cisco did it in his day Google did it in today Netscape didn't today Seems like the right thing to do for our customers our employees and our shareholders well and you're going after some some big markets, right? I mean I presume like most CEOs slash investors you're looking for a big market and you've obviously targeted one I mean you've publicly talked about Oracle EMC as really ripe for a change now of course you're again. There's this co-operative right you're in with ICI They're a strong EMC partners, so there's this sort of ebb and flow of friend and foe But talk about that a little bit in terms of the size of the market that you're going after today We believe we addressed 20 billion dollars manual spent the high IO market and a performance optimized tier 2 market That's today and all that should be in memory and and it's not because we say it's because Larry Ellison says it Oracle should be in memory Joe Tucci himself is tall about the future of his industry But even more importantly the application software vendors Okay, the SAPs of the world who's narrow investor and partner the the oracles of the world the Microsoft's SAS's of the world they're saying all my applications should run in memory and I want persistent memory I want network shareable memory and that's the way you should run your business And so the customers are saying yeah, I should run my business Why should I only take 10% of my data for 90 days of a year when I can have 100% of my data in real time for all 20 years I've been in business You're gonna have a much better business we analyze that kind of data and so the customers are pulling it And what we're doing is we're working to deliver the technology and all the big conglomerates have amazing software. I mean VMWare Greencloth, I mean emcee software assets are amazing Oracle software assets and the best in the world Of course, that's what we do. We make the software run faster where we see competition is usually with a Sub-segment of any company, you know, we're technically competition sub-segment of every company of IBM of HP But in reality, we're not because we're really focusing on a different problem Which is actually to give the end customers the ability to analyze and transact on huge amounts of data in real time Simultaneously, and that's what we do. No one's over done before we had Nino on last week from SAP ventures We were down at Sapphire and we talked about the investment in violin You mentioned SAP in memory, of course, we got a big dose of Hannah Yes, if you're an SAP customer you're going to hear a lot about that When a company like SAP which is corporate venture invest in a company like yours, is there much discussion about okay How do we you know dovetail into things like Hannah or is it more they just see a trend and and want to get a piece of the action? It goes the other way. Usually the corporate investment comes after there's a reason to be in the market together And so really for all of the investment partners that we have that are corporate It's because they saw a match in the technologies that go to market and they really want to support the match So, you know as a small company You need to grow the company be able to embrace and work with the companies as big as an SAP of the world And so the investment is really just a show of support for what's a a natural fit in the marketplace Either technologically or go to market in the case SAP i'd say both. Yeah, it's a wonderful thing for the customers It's a wonderful thing for the vision of the company. It's a very Early on vision you have to give them a tremendous credit for realizing that the software layer itself should be reshaped over the next Seven to ten years to all run in memory and eliminate in the software itself the bottlenecks And so they started that program a couple years ago. We happened to I believe Accelerate that program's ability to go ahead and deal with large data sets at the same time We take all the legacy ERP and we can move it in memory right away Yeah, and that's really exciting and you know our stance on that david florier has basically come out and said all active data will be in Flash or memory, you know eventually So just going on to that So we only have I I got the break sign like three minutes ago, but go ahead I just have a long question which is You you the active memory is going to be uh in in persistent flash type storage Where do you see the value in that chain? The management value where are the breaks in that where you're going to fit in You know the next software companies that are going to come in and take advantage of that The new isv or the new software the ca is etc So I think we work with dozens of of application level be they sass Or be they existing providers of enterprise software that are going to take advantage of that So very clear the application layer the end stuff to touch the customer be an analytic be it a a Salesforce application be it a A big data stack whatever it is Those folks are definitely going to benefit from it. They can write the code cheaper and faster and better In the middle, I think it's an open question Um, certainly violin will have a full set of management features to go ahead and manage our memory spaces into the Hundreds of petabytes into the exabyte layer And then the big middle cat question is Is uh things like the virtualization layer How much is the virtualization layer on you can today run vm inside of island box if you want to I understand the vision emcee house for the future So does the virtualization layer take over a thing does the layer the memory layer hold it? Or does the application as it does in oracle today is the application owner take over don't know the answer It's exciting to find out all right. All right, john. Well, listen. Thanks very much for coming I got a lot more stuff for you I mean, I really would hope you can come back and we can get into sort of your decision to go deep into the you know Your own controller design you guys have added value there You seem to get a very strong technical team You've got an interesting combination of business and technology as your background And uh, so hopefully we can continue this down the road jayden look forward to thanks for your time All right. See you again. All right. Thank you right there. We'll be right back after this brief break Okay, we're back. You're watching a special production of silicon Dot tv's The cube this is the flash flash cube But we're actually bringing you a lot of different content that we've collected excuse me around the industry We're extracting the signal from the noise, which is what we do on the cube Try to bring you the best information that we can this is davilante I'm here live in malboro, massachusetts I'm with my colleagues john furrier and david floyer out in palo alto david floyer We just listened to don basile The ceo of vial in one of the things i i've said is that vial in knows how to sell They've got a lot of sales reps xdmc. They've got you know guy like gary veal running running europe These guys are selling at very high levels of the organization. They're selling business value What do you make of what you just heard from don basile? Oh, he's uh, he's a pragmatist and he's as you say He's a very knowledgeable person about storage in general and and this particular marketplace So he's he is a great ambassador for the flash industry. I particularly liked his He's a his long-term vision of how applications are going to be redesigned in this new environment How that's going to take place? He's talking about the current vendors Readjusting their applications. I i'm i'm not quite so Confident that they will be the standard bearers. I think it's going to be new cloud service providers coming in and Taking over some new space a new white space in those areas and doing things in a different way But but I agree that the application value is where this ends It's it's allowing applications to be more useful more agile more reflecting of the Rapidly changing requirements of a business. That's where the flash the the persistent storage story is going to to pan out John, um, what's your take on basile? You're in the valley basile is don basile is kind of a interesting fellow He's he's got a you know phd. I called him doctor basile and computer science. He's got a business background He's obviously knows how to how to grow companies. What's your what's your take on him? What's the scuttlebutton the valley about don basile? Oh Dave as you know, I've had a chance to sit down with don multiple times get to know him as an individual as an entrepreneur And as the CEO of a pre-publicly traded company and the thing about basile that i'm really impressed with is He is a maverick and I've called him that on the cube before because he's a hard What I call from you know, he's coast term hard charging Individual he was at stanford has a phd. He is smart. Okay. He understands kind of what's happening at a root level He understands what's going on at a macro level. He's had his hands at some startups He does a lot of investing he is plugged in and I think he is a the maestro of orchestration And competitive as hell So, you know the thing about basile that I think is interesting as a CEO on this day in day and age is He has that 20 mile stare that he can look out in the marketplace and see the vision as david was pointing out I think that systems vision looking at the application model That's moving quickly with software rewriting into new architect is using persistent memory both in memory and using flash He sees it So I think it's pretty obvious. He's got the 20 mile stare But he also could orchestrate things within that system and also he's competitive So, you know hard charging smart super competitive and the word in the valley is he's a he's a great guy to work with You know, he's but you know, he's running fast if you can't keep up with him He's going to run right by you. So I think I like his competitiveness. I like his technical and I like his business savvy So one of the things that we talk about a lot, of course in silicon angle wikibon is big data Not you know, before I get into that, I want to just take a moment wikibon.org is where all the Research is go there. It's all free. It's peers Contributing it's it's sharing information and ideas Silicon angle.com is the reference point of tech innovation So go there for all the daily news and analysis Check out those two sites and and we really appreciate you watching But one of the things we talk about a lot in silicon angle is big data and wikibon If you go to wikibon.org slash big data, you'll see a compendium of all the research We've done the best of big data Now john and david we talk about big data a lot. We talk about open source We talk about hyperscale the confluence of these big trends is coming together and last week in he chose uh, who's the vice president of Of products and strategy at ibm's big data and information management business Join me at the 411 event. So I want you guys to watch this video from in he cho An up-and-comer at ibm very articulate. So let's watch this video and we'll come back and riff on it Hi, this is davilante and we're back. We're at 590 Madison avenue And we have a cube alum in he chose is here and in he thanks very much for spending some time It's always great to see you. Oh good seeing you, Dave. Yeah, so we saw each other last week in california The big data big data management announcement. Oh, yeah, it's very exciting We've been covering that like crazy on silicon angle and wikibon So take us through what you guys announced and what customers are saying about it. Oh We're really excited and the feedback has been tremendous since the announcement last week wednesday It's really about speed and exploration big data at the speed of which businesses are operating And so what we're doing is we're really enabling clients to consume it better and more So one of the big things that we announced was big sequel capabilities and our infos here big insights Because we found out, you know what clients were having a hard time adopting it They didn't really have the skills in house to get this stuff set up quick enough That was one piece another piece of the consumability. We also announced the ibm pure data system for hadoop And that is in an appliance like simplicity so that it's really a data load ready system for big data Projects and within under four hours. I mean, that's what we're committing to clients. So a lot around Consumability performance speed exploration Yes, so we just had steve mills on and we've been talking about the synergies across his lines of business So it's interesting to see you here as a big data analytics, you know person Supporting the flash announcement talk about that a little bit. Yeah IBM is really investing in flash at an incredible level So we announced today our strategy around one billion dollars of Investments in research which includes development acquisitions as well as other Optimizations that we'll do both in the hardware storage as well as software Layers the other aspect which is exciting is this kind of marries into some of the announcements that we made around db2 With blue acceleration. So blue acceleration is our new dynamic in memory capabilities which allows you to take advantages of things like, you know active compression So now you can have more compressed data that can take advantage of kind of the advantages of flash or taking advantages of more parallel computing Capabilities and some of the hardware trends that you're seeing in flash and systems design So let's talk about a little bit about the big data space. You're obviously, you know, really immersed in that area IBM has just done a tremendous job Taking its analytics business and really driving to what we've cited as a number one Status in the industry. So What's going on there? You're seeing the sort of old world db2 and the new world a dupe come together You're seeing real time in memory. Take us through help us squint through all the you know machinations in that business What's going on? Well, you know what it is is It's actually sort of a confluence of things coming together, right? Clients are saying, you know what they're new types of mobile applications Some clients want to access things via the cloud There's higher degrees of power shifting to consumers And all of that is try driving kind of fundamental changes in how we do entire systems architecture design and software design Now if you think about it from a data standpoint Really what clients want to do is just really get better insights faster And the investments that we're making are enabling clients to do that and some of the new technologies And the ability of bringing all data together with the new technologies is much more affordable than it's ever been And it's just kind of an exciting time to get clients all started and and some of the use cases that we're discovering So one of the big themes that you hear is bringing real time to hadoop And it's there's an interesting debate going on. I was talking to stefan last week at the announcement stefan from datamere Who's who's he's an early day practitioner of one of the uh predecessors for uh Yeah, he'll build the build out there. So and he's essentially saying you know what hadoop is is really designed to be batch And and you know the the rest of the world, you know, we'll we'll connect to it Um, but a lot of people disagree with that a lot of people are really Trying to make hadoop look more real time. What's your take on all this? You know, uh Well, I have a very strong point of view on this because that's why i'm asking you I put you right in the spot because I Your angles well, you know what real time is is all relevant right relevant to the type of workloads and types of transactions That you want to run and types of analytics you want to run so I've been really the only vendor that has what we call streaming technology the ability to consume All sources and all types of data in true real time So and when you think about hadoop the capabilities of hadoop is really offline batch, right? And and the design point is to leverage sort of the distributed in memory and file system capabilities of that kind of architecture With stream computing it's a completely different design of being able to sense and respond I mean when I think about hadoop think about kind of the things that you do in your head, right? You're in a mode of deep reflection You want to sit back think about think about what happened during the day think about what happened during the years Versus in your in sense and respond mode If something happens the last thing you're doing is actually thinking Deeply at that moment. You're naturally responding to based on instinct prior experience All under less than a second So those are kind of very different Requirements in terms of their architectural design So I think pushing that in terms of the hadoop system is going to be a real stretch Streams is digging available. So that'll allow you to process data in real time before you even persisted, right? Oh, yeah, absolutely It's it's an ability to actually Stream the data ingest the data as it's coming out Understand kind of the anomalies of what's happening and then take action very quickly And what's exciting about is we're also enabling things like being able to take The modeling work you do in spss those operators could actually reside in stream So even as the data flows in you can optimize even the models in which you're predicting and analyzing some of the behaviors and patterns And then optimize that into and and automate that into your workflow and process Yeah, you guys have a lot of the pieces. I mean you've got the cognose piece You mentioned spss. You've got the informer's piece, which gives you time series You've got your own Hadoop distribution now and big insights streams. I mean Hi, this is Dave Vellante and we're back. We're at 590 Madison Avenue and we have a cube alum in he chose is here And in he thanks very much for spending some time. It's always great to see you. Oh good seeing you, Dave Yeah, so we saw each other last week in california the big data big data management announcement Oh, yeah, it was very exciting. We've been covering that like crazy on silicon angle and wiki bond So take us through what you guys announced and what customers are saying about it Oh, uh, it we're really excited and the feedback has been tremendous since the announcement last week wednesday Um, it's really about speed and exploration big data at the speed of which businesses are operating And so what we're doing is we're really enabling clients to consume it better and more So one of the big things that we announced was big sequel capabilities and our infos here big insights Because we found out, you know what clients were having a hard time adopting it They didn't really have the skills in house to get this stuff set up quick enough That was one piece another piece of the consumability. We also announced the ibium pure data system for hadoop And that is in an appliance like simplicity So that it's really a data load ready system for big data projects within under four hours I mean, that's what we're committing to clients. So a lot around Consumability performance speed exploration. Yeah, so we just had steve mills on and we've been talking about the synergies across His lines of business. So it's interesting to see you here as a big data analytics You know person supporting the flash announcement talk about that a little bit Yeah, ibium is a really investing in flash at an incredible level So we announced today our strategy around one billion dollars of Investments in research which encodes development acquisitions as well as other Optimizations that we'll do both in the hardware storage as well as software Layers the other aspect which is exciting is this kind of marries into some of the announcements that we made around db2 With blue acceleration. So blue acceleration is our new dynamic in memory Capabilities, which allows you to take advantages of things like, you know active compression So now you can have more compressed data that can take advantage of kind of the advantages of flash or taking advantages of more parallel computing Capabilities and some of the hardware trends that you're seeing in flash and systems design So let's talk about a little bit about the big data space. You're obviously, you know Really immersed in that area. Um, ibium has just done a tremendous job Taking its analytics business and really driving to what we've Okay, we're back. This is Dave Vellante. I'm here with david floyer. Who's on the west coast right now david Before we queue up vincent sue who's the cto of ibium storage division. I want to talk to you a little bit about of this notion of Getting rid of the horrible storage stack. So What do we mean by that? What let's talk about atomic rights. Let me start there david. What are atomic rights? What does that mean? so an Atomic right is a right that you can do once and once only um, and the normal way that rights are written using the scuzzy stack is that you have a sort Two phase or even three or four phase commit where you go backwards and forwards. Remember that journey to uh To from san francisco to los angeles you you queue in every now and again and say, okay How are you doing? Have you got it? And you acknowledge it back and then say something else So you you have to make sure you know where that where that stuff is and then be able to Make sure that it's safe the other at the at its destination Atomic rights are single rights directly to the storage media in this case flash and the way that it's done is to Write it to the controller of the flash and there usually there's some sort of capability for Persistent capacitance for example on to d ram or using m ram At that moment the data is secure So we can ensure that the data is absolutely captured and that can be done in around 100 nanoseconds So an atomic right really changes the name of the game Instead of you talking about milliseconds or even microseconds through the to to the traditional ssd You're you're able to get it down into these nanosecond areas 100 nanosecond area So the the benefit of that from a database point of view is i've committed my right I know i've got it safe now i can Process the next transaction and i improved my Rate my ability to do database calls immensely as a result of that Okay, so who is doing atomic rights? I mean obviously fusion i o was really the first to get to market with that, correct? Yes fusion i o is the only one that has put together all of the software to do this and the software includes a virtualized Vsl layer that they describe it It is the file system In their case direct fs and it's the apis that can be used to to to run this So they've put together the whole stack and and in fact they've also committed. It's all part of linux It the only people from a database person using it at the moment are my sql So it's a limited but actually in the area that they're going after the hyperscale environment that is the environment that That's that's the software environment linux and my sql environment that people like facebook are working in So it's uh from an enterprise point of view It's a long way off doing it on db2 or oracle But it's a very important first step of creating this total stack and clearly ibm and emc Have to also create similar stacks To be in the marketplace hp as well They have to create these sort of stacks. There is a there is a industry standard nvme That's being talked about but that year hasn't yet got down to the level of atomic rights Yes, so why do they why do these other companies have to have to develop their own? Why don't they just write to nvme? Well, it's not yet at a level of being able to write these It isn't defined in that stack yet as an atomic right. That's a sort of future dimension of that They it's a very loose nvm. It's non-volatile memory and they're trying to do this for the next generation of non-volatile, you know, Memoristers or whatever comes after so it's a long term ambition And what fusion ios is a much shorter pragmatic. Let's let's get something out there and let's iterate on that All right, so you're saying that you're saying nvme is not ready and companies like ibm and emc and maybe even vmware Shouldn't wait. They're actually going to have to go develop their own proprietary standards In particular vmware vmware hypervisors can take immediate advantage of this sort of technology And build it into the to the ios stack which is always the weakness of virtualization. So vmware hypervisors both from microsoft and from And from vmware should should be investing in this very heavily and I would think that ibm as well I mean ibm's got hypervisors ibm's got file systems ibm's got os's ibm's got flash They got server expertise. I would think that within the ibm stack ibm could really make things like db2 Uh and and other servers linux servers, etc run really really fast ibm at his heart is a technology company and it's a systems technology company. It's always Provided that systems view of the world and that's the opportunity that ibm really has is to bring together Under steve mills all the areas in the development tools All the areas in linux the open systems tools all the areas in the in the Servers themselves bringing those servers close creating that those caches coherent caches On the clusters of servers Bringing together the storage with the flash bringing together the software infrastructure to to to enable this This is right in their wheelhouse and ibm del hp ibm and hp are More traditional systems vendors those are the people that have those capabilities and It's a surprise to me for example that I hear that ibm Maybe trying to sell that to that part of their server division. Well, what about oracle? You know, where do they fit in this whole thing? Is this something that they you expect oracle to do as well? This is the atomic right capability and absolutely they they should be they they Have flashed they have a lot of information about flashed. They know about the flash. They certainly are a systems Software supplier. They got database They definitely got a database or two They got the the world's leading database. The interesting thing is I mean the complexity of those databases to overcome How to write data from here and wait till you get back from los angeles the complexity of those databases is immense as things gets Flash becomes more and more available and atomic rights become more available So it's going to be interesting to see a lot of other Database vendors come into the marketplace and I've never seen the interest in database that there is at the moment it's a burgeoning here in the valley and It's if that is if that is nothing else That's a wake-up call to oracle that they really need to get on board And get to understand what this cloud and these cloud service providers are looking for Yeah, as john furrier said database used to be boring and now it's very exciting. All right, david I know you have to run to another call That was a good setup for this this geek out segment that we had Down in new york with vince and sue who's the cto of ibm storage division So let's let's what we're going to do is we're going to watch this video and then we're going to come back And then john furrier and I are going to pick it up david get back as soon as you can Then we've got gary orenstein coming in from fusion io and scott deetson who's the ceo of pure But right now let's watch vince and sue of ibm keep it right there Okay, we're back. This is dav valante. We're here at 590 madison avenue in new york city at the big ibm announcement today Flash ahead i'm here with vincent sue who's an ibm fellow and cto of the storage business vincent Welcome back to the cube Good to see you again. So uh big announcement today You guys are all in on flash talk about flash as a game changer. Why is it such a game changer from a technical perspective? You know over the last, you know 50 years The whole system design and the software applications are written pretty much to start to accommodate the io problem As you can see that for the last, you know 10 years the cpu has improved 10 times they were improved 100 times but storage three performance was only improved 1.5 1.2 times A lot of the system architectures and software are really trying to work around their problems Now when the io bottleneck is Removed okay use all of a sudden a lot of things become possible Okay, not just the traditional application now you can do more because we can do more with a much shorter time Yeah, so um On the one hand we heard from customers today that they're basically dropping in an all flash array It's a block based device looks like any block based device and it immediately accelerates performance on the other hand over time There's great potential to sort of re architect infrastructure and applications What do you see is is is that potential and how will that change infrastructure applications and ultimately business? Yeah So for example today The Let me just give you a most one of the most obvious example today in the database design We everybody do database reorganization database re or right you have to re or database Why because you want to get the best performance? So you need the data sort of adjacent to each other On the on the plan Okay We need to question those business probably not doesn't need to do it anymore or doesn't need to do as often anymore So all these parameters that we do Tune our system those had to be rethink Even in our io path. Okay today we try to buffer up as much as possible Because go to this is very slow right how much metadata you can put in the system to make the sense So allow you to make sense from the data Those things in the past is very hard to do because every little io go to this is very slow But now because the flash flash enable system allow you to put a much better metadata reach systems available So you can do a lot more with what you have So let's talk about metadata a little bit. So metadata today sort of locked in The the device whether it's a network device a server an array And it's sort of controlled by by that device. Do you see that changing? Will this whole notion of software led or software defined, you know infrastructure Allow metadata to actually be a shared resource First of all first question and second question is where is that going to get managed? Okay, so you are absolutely correct that the metadata is where First of all, I mean metadata will allow you to make sense from the data from the raw data you have So that is I think that we are getting to the a we are getting to the point that this world going to have a very very metadata reach informations Okay, but in the past because the metadata is you know compared to the raw data Minutas are much smaller, you know, you go to those the smaller io It costs you a lot So a lot of people that buffer up the metadata in the in the DRAM memory So it's sort of limited amount of data metadata you can have but these days was the Now that was the flash systems available You're able to do a much better reach metadata and to be able to share between the different applications The thing is that these days they are the a lot of time we see over and over again Different applications have the same content But we they cannot really share the data without reshape their metadata around the raw data So now with this kind of technology will enable you to be able to share those Uh, share those content much easier once you can transform those metadata quicker Yes, so you heard in my in my panel that I was hosting I was doing some back of the napkin calculations last night and I had I had, uh, you know CPU speeds at nanoseconds 10 to the minus nine and and and disk speeds 10 to the minus three six orders of magnitude Delta I'm not sure if that's exactly right, but it's it's very close enough, right? So my question then is who's going to control that metadata? You can't really control that metadata from slow storage. Doesn't it have to be controlled from fast servers? In the past people try to put automated in the DRAMs Because if you control a metadata in a slow storage, then it's just too slow, right? It's just especially the metadata is a very small IO small IO is extremely expensive to the hard disk But the problem is if you want to have a large amount of metadata to make your data To be able to make sense from your data that you cannot put everything in DRAM And that's where the flash or flash system coming to play here. Excellent. So, um, so how do you see this thing progressing? You guys are putting a billion dollars in obviously you've got, you know, investments organic and potentially inorganic We're going to ask Steve Mills about that. But where do you see this whole flash thing playing out over time? Okay, so the flash optimization is across all the layers from the fundamental core technology. That's what The Texas memory system has done a superb job in the you know to make the flash more Durables and you know high performance Then you have we have an overall system design the power system Z system the storage systems. How do I design a system? So rest of the system is not going to be the bottleneck, right? Then we have a middleware design to take advantage of that. Remember I talked about the database reorgs And there are a lot of you know, the software defines storage interface to allow The allow the middlewares application to be able to truly take advantage of those technology Because at the end of the day people don't want to spend their time defining where the data placement going to be They want to be able to talk to this This this this devices at the end of the day what ibn vision going to be Take this storage problem from application all you see is Infinite flat space you can just rewrite to it and everything will be persistent back to the days of single level store And it's uh, it's super fast. All right, vincent sue Thanks very much for coming back on the cube and great to hear your perspective So keep it right there. Everybody will be right back with our next guest