 Live from San Francisco extracting the signal from the noise. It's the queue Covering Oracle open world 2015 Brought to you by Oracle Now your hosts John Furrier and Dave Vellante Okay, welcome back everyone. We are live here at Oracle open world for Wednesday This is day three day forward event how you look at a day one was Monday day zero Sunday This is the queue silicon angles flagship program. We go out to the events and extract the symptom noise I'm John Furrier the founder. I was looking at my coach Dave Vellante founder wikibond.com Next is Mike workman senior vice president of flash storage systems at Oracle welcome back to the queue. Thank you Great to see you. It's good fun. So today was under the hood day down a telly on stage Fowler This is where all the the geeks get their inner geek going because there's a lot of hardware software Action going on and there's a huge shift in the market. I mean Dave pretty much painted the picture Hey, we're in a new era client service kind of dead or over We're gonna build on top of that Flash is a big part of that. Yeah, but now you got in memory. Is that kind of how's that gonna work together? Can you just tease that out for us the future of flash in general? Well, I mean flash is here to stay and tell something's better You know, I mean 3d cross-point. What's next after that? There's a lot of memory technologies out there I think that the key thing is is that we seem to never really go back in other words You know once if you get used to five-minute service and somebody Serves you in six that's unacceptable and when people get hooked on flash and the latency Difference because it's so dramatically different and the way it can accelerate apps There's no going back. I love that getting hooked on flash. That's definitely a cube gem moment I mean people are people are getting hooked on flash, right? I mean talk about that because that really is the headline people are hooked on flash And they want more and what so what is that dynamic of why they hooked on flash on the latency? What's specific things beyond that the benefits that they get on that and then what's the next high for them? Is it is it in memory? So what's that next addiction? I mean because you're right You're a hundred percent right on that people get hooked on it. Well, I think you know people we don't wait well I mean honestly, you know what no matter what society global society. We don't wait well And the when you're when you go away from flash once you're used to it You're waiting a long time for things. I mean dramatically different, right? And so it changes your expectations and the expectations are what what we buy to we buy to fulfill expectations If we're in building a it data center to either serve our internal customers or the world, you know our external So is disc the new tape or do you think the wikibon forecast are a little too aggressive? They're too aggressive and I don't even know what they are give us some No, I mean, I'm telling you there, you know, you can't count out disc for where the petabytes that it's going to serve It'll be 10 to 1 right if you look at Everybody has tried to argue this. I'm of my product is a flash storage system Right, we build I think the the fastest flash storage system in the industry and it was built for oracle by oracle So it might be good for oracle because we built it But that I think that the key Thing of it is is that no matter how you look at it. I couldn't procure enough flash To store data on the data that exists today and the data that will exist next year. There's not enough Money for fabs. It's true. Two trillion dollar A ticket by some estimates to build fabs to supplant hgd. So that will not happen So this would be a supply problem if if in fact a massive and if you started today It takes how long to build a fab two years. That's not happening Yeah, so At the capacity level that they need it's two years for every fab times all the demand that's required Yeah, two trillion dollar check I mean So what does that mean he doesn't have that? Yeah That's a cell as island u.s. Government, maybe They don't have it either It's mid-borrowed So so what does that mean for flash pricing than than mike? How should we think about that? I think it's gonna be You know, it'll continue to decline because everybody wants to sell and there's competitors, right? But I don't think it's gonna be a freefall of price because there's more than enough demand to get you know to Make the the squires happy But I think that also for hgd folks it spells well for them too because there's only two suppliers of a big volume left And people are going to need it so But there's things you can do with flash obviously besides the fact that it's not spinning that you can't do with Amazing. So I wonder if we could talk about some of that. I mean the the rate that you can, you know Compressor d-dupe the way you can share flash out of a single copy. How is that affecting business outcomes? Flash changes the whole game And the reason you get hooked on it is because applications When when you If you give somebody Uh 50 petabytes of free storage they somehow figure out how to use it, right always and all of a sudden it becomes a requirement Right. I mean, that's the way it works that way. You deliver somebody a component and all of a sudden they build Solutions around it. Well, you can build things around flash that you cannot build You take a storage system that can do two or three or 400 000 iops How many uh cloud customers can you put on that like in the on our fs one a lot? You can't support that with an old-fashioned architecture using hgd You can't get the density of customers per platform which translates the cost right and performance and expectations and latency and all of that So you get hooked on The speed and the things that it can provide you and then the ecosystem starts building things That require it So that's why I say you can't go back because the ecosystem now assumes that that latency is there and it's it's getting more and more In our product. We did something. I think really interesting We made it not only good at what you expect from flash, which is high i o And low latency, right? We made it really good at sequential streaming Really good and people say would you do that for a video? No And not for video you don't need that but what you need it for is when you're running a huge ol tp kind of a environment A hundred oracle databases all pounding away and you need ol tp You also have to run backups while that's running and you can't run backups at 600 megabytes per second one big stream six gig or 12 gig Per second those are the kind of numbers you got to be talking about so we did that whereas the other guys are way I'm glad you said that because Flash changes backup that the way we do backup today can't survive right flash Right I got I got to ask you on the flash name because there's two things that I see happening one is You know, they're going to use that. I agree. They're going to use up that space. It's either going to be Hey, we got a store stuff. I'm lazy. I'm just going to put some stuff use that flash put some stuff They'll figure it out later or it's really a mission critical app They'll use it either way ones like giving the teenage kids the car keys And you drive it too fast and you know, they wreck the car and the other one is just I'm I got to move fast And I got a lot of technical requirements systems requirements. So that brings up. Okay. They're going fast On the deployment with the performance. They hooked on it. It could be intoxicated. It could be legitimate requirements Security becomes an issue because now if I'm just going to be using flash and just throwing stuff out that this I might not be careful enough to say hey What's going on with my data? So how do you guys address the security question in the in the fs? Well, all this going on. I mean, I don't say lazy, but people are throwing stuff in flash You know, I don't want to I'll get to it later. I'll manage it later And then there's real requirements, but security has that fit in that I think it's a fair question and I think larry's done a good job I imagined david an excellent job of talking about security and the oracle software stack But you know, let's be honest. It needs to be everywhere And as larry said on sunday night, it needs to be at the lowest part of your system You can't leave it Somewhere for for fear of of it not getting done correctly. It has to be everywhere distributed and even at the lowest part So in our oracle fs system We encrypt the data on the drive Now you can use it with oracle database transparent data encryption to You can use it with other applications that encrypt hire up the stack fine But if you don't it doesn't matter they can't steal your data arrest because it's encrypted on the drive And it does so with zero performance loss Most people you will explain that explain that's a huge issue. You come out encryption now, right? Okay, so What happens is is when when when people talk about encryption in general our competitors talk about Well, if you turn it on we can encrypt your data for you. The problem is you can't de-dupe it then It's encrypted and another problem is they'll never show your the benchmarks or the performance metrics with encryption on Because it sucks cores and software We do ours in hardware cure hardware distributed in parallel and no one sees it It's transparent if you turn it on or off you can't tell by looking at the performance Which is an awesome feature and it's part of what larry said. We're not the de-dupe piece. You guys can de-dupe it You can't de-dupe with encryption. Okay, so that's off the table. That's off the table What we do instead of of de-dupe like for oracle database We use htc and htc is really powerful layers of compression Depending on how deep you want to go and how hard you want to compact it and how much work you want to spend Rehydrating it the database cpus can do that and it cuts down network bandwidth and capacity that you use but it's we're we're oracle And when when you run an oracle database, we trust them to know what needs to be encrypted and what how it should Be encrypted that so they can turn off the encryption in our system if they want to But you don't have to you can leave it on it's sort of double All right, so I got to ask you I got to ask you the cube question. We've been asking everyone Dave addressed it on his keynote We'd love to look at the big changes, but you know the management team in oracle is pretty solid right now very solid great great team Dell now has acquired a team with emc, but yet they're in they're in a transition phase With all this landscape changes going on Storage is still hot with seeing seeing a lot of startups out startups out there storage startups What's your take on this innovation? Was it's going to be flash in the pan? Will there be viable players out there and because you guys are building some a pretty high fence from a differentiation standpoint Talk about that difference your differentiation Inimidability versus the new guys in the blocks coming in and the changes at dell emc if you want to comment on that Yeah, that's a that's a great question and it probably deserves more time than we have. I'll try to Cover the highlights First of all, I ran a startup for 10 years and if there's such a thing as a 10-year startup Before you are your first customer. Yeah, Michael tell says he's running a startup. No, no Definitions, but um, it's hard to to to fit yourself in the ecosystem of the whole software stack Anywhere if you're just doing a component right and the component manufacturers for flash storage systems that really mattered in my opinion I don't mean to be mean they've all been bought by the big players who are going to attempt to build a End-to-end solution But if you look at the the layout the landscape today the only Real people that have a I mean to me Oracle sort of stands head and shoulders above everybody else in terms of having all the pieces I mean Larry's been on this for 10 years people but You know as usual some people said he was nuts, you know He's he's actually investing in hardware. He's he's building microprocessors and operating systems and Switching gear and all this stuff, right? Why why everyone else is divesting of these components, right? And he's building them the old man's lost that he just loves tinkering but but you know, I never bet against this guy, you know, he's got a plan because what he sees is that People don't have the time or the inclination at the rate of change the way things move to build piece by piece anymore You can't you know who has the time? I can't even hardly put together a home stereo system by myself I mean, I I know what i'm doing. I used to design that stuff, but for god's sake it's pieces and work and I team managers don't want to become this anymore that it's the old best of breed So what he's building is the best of breed integrated together Seamless between the cloud and the on-premises solution to make them look transparently You know the same whichever one you elect I mean all of that stuff going at that as a component manufacturer is tough Yeah, so let's take like a pure storage. They got women publics. They got a valuation and then these other stars Are they then they're building just a feature or they say best of breed tactical solution And it's you know, and it was brilliant for them. So i'm not I don't I don't want to throw rocks at people A great little great big business now or you know valuation was fantastic for scott eatson, right? You love to see that we you're here at silicon valley But can guys like that is you reach is a breakout velocity? I don't know the problem is well, I mean you've got a ways to go, right? I mean unless you consider that the ipo, but if you if you want a profit Well, no if you're spending two to make one that's something's got to give so you got to get that breakout velocity Yeah, I I think I mean the real problem is is that people everybody's made their play And that's a tactical solution as it becomes integrated as we've done at oracle into your whole stack and you've already got it You know, there's a lot of disadvantage to going and sort of disintegrating And going and buying some other solution and trying to fit it into the whole puzzle. It's just a flash We all we know it changes everything, but it just Doesn't seem as disruptive as everybody thought it would be. I mean do you feel disrupted oracle because of flash? No, you're embracing it I don't think flash is the reason they're getting disrupted. There's other factors there. It's more cloud, right? So Cloud's disruptive. I think everybody even though they started working at they weren't all sure what they were working on Just say cloud let's keep saying it Let's play musical chairs. So in the flash business, there's a lot of consolidations going on What to your point earlier when the music stops If you're just a component and the music stops and you're not in a chair Being bought by somebody else or a fortified market position You're going to be left standing and you're done, right? So what are those chairs? What's the musical chairs? Model going on and flash right now because obviously the integrated stack makes a lot of sense So amazon doing with their stack. You guys have a here integrated stack platform and integration as a key part of it engineered systems We like it. But where's what's the chairs like? What's the music the music will stop? I I believe there'll be a scale point where The components some music never stops That's one thing I think pretty the chairs may change the chairs may pull up But what chairs are left? There's a lot of people walking around the chairs right now There's there's more consult. There's there's fewer and fewer chairs. I think right there's consolidation going on We know you you see it every day and it's got to be fascinating. It is to me Yeah, we love it. It's great for our business. But the music doesn't ever stop. But I mean like well, there's the cloud chair There's the integration chair. Yeah, there's the commodity chair Well, let's let's do it another way if you're an integration chair and you don't have a cloud You don't have an integration chair I don't have a chair right and I think uh, you know, Dave mark herds made that point Very clearly that if you're a public IT company and you're in you you want to be here 10 years from now If you're not in the cloud you're done That's an interesting thing to think about because how many companies out there Maybe just killed their cloud How many companies out there aren't thinking that way they're trying to survive in the old model And that chair is going to go away. You don't have a cloud. You've got to play in the commodity chair Right and that's isn't that what michael del's doing. He's saying all right We don't really have a cloud now. I don't know forget emc for a second The strategy for del was always sell to the cloud. Okay, and make 19 gross margins used to be in the disk drive business You make a little bit of money there Let's say strategy Okay, that's not oracle strategy, but it's a strategy You know oracle strategy is to Integrate up the value up the stack. It's not volume top the bottom, baby Right. Yeah, so amazon. We know what this strategy is google microsoft can compete with that You're not trying to take amazon head on even though you'll price to amazon head on On some aspect, yeah, right? Yeah, but the strategy is not to go like this head on with amazon amazon per se we're trying to to be the low-cost supplier I mean larry's made that clear of last year's oracle world for infrastructure because you can't you can't do it Unless you're you aim at that you may not have to hit it exactly But you should try and if you you know, you should try to be the best Okay, but so what if you didn't have database and applications and middleware that would not be as interesting as strategy No, it'd be horrible. I The integrated means all of the stuff that people want to run. It's about the application at the end of the day nobody I mean i'm like I supply infrastructure, right? They just want it to run. They want it to be secure. They want it to be Reliable they want it to be really really fast Because that's part of the cost equation as everybody has pointed out They want all of that stuff, but people don't buy A disk drive or a flash drive. They don't care about that. They care about the it results, right? They care about the applications They care about running their business and we've said this for years But the thing of it is is it's more and more true every day if you if Nobody knows what's in this and most people don't care And it works it works better than the other and that's the whole thing They just don't care. They want it to work and they want to get what they get They want to know how to get over to the end of howard street. That's what they want On our post keynote wrap up that we've got syndicate on oracle.com Dave made the comment you're either car maker or car dealer quoting scott mcnealy's famous line And then one of the analysts danis howl has said yeah oracles now the tesla Which kind of brings up a good point through our cars older cars and then game changing cars It looks like a car but it's different and it's winning. Tesla's winning. Yeah So in a way, I feel like this For oracle is the kind of tesla model because new components is reconfigured engineered differently and runs better Safer more economical people love it sports car. How would you just plug it in? How would you describe if it was a tesla? What are you guys doing? That's more tesla like That's different than the old way. How would you summarize it? Well in the case of flash stores There's less moving parts, you know, we got rid of some of the motors I I don't I don't know Actually go down to the base of the elevator. We have a private cloud Appliance right running with an fs our flash storage system It's sitting there and if you walk behind those boxes You can close your eyes and you can identify the flash storage system And the reason is because you could just feel the cool air blowing over you relative to the heat Generated by some of the monsters we're sitting next to It's amazing power security, whatever. It's all awesome Mike workers senior vice versa. Thanks for coming on sharing your insight and color into the into the flash I'll give you the final word over the next 12 to 18 months What kind of movement in the flash business are we going to see my technology standpoint? What's the big trend that you will be riding? I think the latency push is going to go faster and faster harder and harder towards lower latency And there'll be more and more options and tiers within the flash tier Memory tier whether they be alternatives provided by intel or others that allow us to get to latencies that are Normally only seen in direct attached server storage And I think that that's going to happen, right? There's also going to be an acceptance that it might be better To have this all flash tactical solution, which we have a product the all flash fs one or To also have same software same api same control everything the ability to also Yes, take advantage of disk in your data center because it's a reality and you're going to need it Why would you want two products to manage it? And I think people stop the once they realize you are in the 100 microsecond latency phase They'll stop the complaining about the fact that they see a disk somewhere in your data center If you need it, it's a tool you use it, right? And I think that that's that reality is going to shrink it to sink into everybody Security is definitely a thing and and we'll see that I I uh that would be more of the same the chairs Keep moving though Never stop but there the chairs are moving away based off for some Mike workman as always Very colorful very technical great to have you on the cube. Thanks for sharing your insight with us This is fantastic. This is the cube live on howard street where we shut the streets down actually not us Oracle does The 60 000 attendees we are rocking it here inside the cube. We write back with more after this short break