 Live from the Moscone Convention Center in San Francisco, California, it's The Cube at Oracle OpenWorld 2014. Brought to you by headline sponsor QLogic with support from HGST. violin memory and MarkLogic. Now, here is your host, Dave Vellante. Hi buddy, we're back, this is The Cube and we're here live at Oracle OpenWorld. Live in the QLogic booth, this is our fifth year at OpenWorld. Thank you to QLogic and our other sponsors. QLogic gives us about two thirds of its booth to run this production here. It's been a fantastic partnership with QLogic over the last several years. David Floyer is here, he's the CTO of Wikibon. He is an expert in a lot of things, IT, data center, certainly the server markets. We forced him many years ago to start looking at other infrastructure like storage. Knows a lot about database. We're in the heart of the database land. David, welcome to The Cube. Always a pleasure to have you on. Thanks very much, Dave. So, you know, the big news Sunday in the infrastructure side of the world, so much to talk about, but let's just get right to it. FS1, Oracle on its last conference call, SAFRACAT said that San and Larry Ellison, echo this, San was down, but we got some new announcements coming up that's going to potentially address that. That was the FS1, which was the sort of original pillar data axiom, redone, now called the FS1. We confirmed with Mike Workman it is actually a new operating system. It's a Linux on Oracle OS. Six full bit operating system. Using the features of pillar data, which were three par like, friends at three par cringing right now, going, oh no way, but similar in that, that was born in that virtualization era, focused on simplifying, focused on quality of service, some of those features they brought along, but I joked with Mike Workman, hey, you look good having been tied up in an R&D lab for the last two years, but that's essentially what happened. Larry Ellison bought the company and said, okay, here's a bunch of cash, let's do the roadmap. Well, let's do the roadmap, here's a bunch of cash to get it done, get it done. And I'm sure it was challenging given what the product ultimately became. I called it the chameleon. So, give us your breakdown, bumper sticker, summarized, FS1, how should we think about it? So, FS1 is a good product. They really brought the best of pillar and the things that I really like about from the pillar are the quality of service and the tiering and they put that together and called it QOS Plus. And it is, they have very fine grain, breaking down of the elements that they can move around they have the four tiers and they have two flash tiers which is the first time or a compelling had a sort of two tier flash, but two flash tiers and two disk tiers. So, performance flash, capacity flash, performance disk and capacity disk. And they manage this quality of service extremely well. What they can do is take each IO and then give it a priority. So, and they will then bump, instead of just taking them one at a time, they will bump the IO's from the higher workloads up and give them a better service. So, very nice design from a utilization point of view, very nice design from a performance point of view, SLA point of view. And they've thrown in a few other things as well which are very nice. They've got automated quality, so automated provisioning. So, they've predefined how you should set up Oracle databases and Oracle applications and even some non-Oracle ones like Microsoft Exchange. And they've done all this work and they can automate this, a single click to automate it. So, they've done a really good job of making it work with Oracle, work with Oracle software. It's a good fit with Oracle software. It's done a very, very good job of bringing it out. I think in that timescale, it's been a pretty good time. Look at, I mean, it's basically replacing the Sun Sand product. Right. Which, you know, I mean, it wasn't winning industry awards, you know, no offense to my former friends at Sun. Interesting, they kept the Sun brand. Yes. I noticed that on the, I'm looking at one of the boxes here. So, we had EMC on yesterday and they said it's a hybrid array, called it a hybrid. Mike Workman said, no, no, it's not a hybrid. I don't know, maybe not to get academic into definitions, but is it, in your view, a hybrid array? Well, it is a half flash and half disk and it does combine the two pretty well. But you can use it as a flash only. However, if you're going to use it as a flash only, you would want to have some of the features that the flash only people have put a lot of effort into. So for example, if you're going to have it as a general purpose flash only device, as opposed to an Oracle flash only device, you would want compression, you would want deduplication to be on those boxes because that's going to cut your price down by a good factor of four or five. So, if you call it a flash only, it's not yet in the marketplace, but for the Oracle marketplace, where you're using, for example, the hybrid columnar compression, that will then work very well with the flash side of it and you'll get very good results. So, let me make sure I understand this. So you said it's part flash, part spinning disk, but I think of a hybrid array as sort of an array that puts some things on flash, moves them to spinning disk. It's not of that ilk. So, okay, but if I am allocating, do I have to manually allocate a volume to flash? That's how it works. That's exactly the same, yes. And it's tiers. Well, I guess if you're... It's exactly the same as the tiered system from Vmax, if you like, except that there you, well, it's very similar. You have the multiple tiers. They have four specific tiers and it'll take it up and promote it to the tier. Okay, so it's got an automated tiering capability. Absolutely, yes. So why isn't it a hybrid? Well, because the hybrid is where usually, for example, if you talk about Tintry or some of the other ones, it comes straight into the flash and then goes down from there, okay? So it's a flash first? It's a flash first. It's a flash first. And this one. Here you have a volume and it'll decide whether it puts it on the flash first or on the lowest volume. Although not all hybrids are flash first. I mean... Well, yes, but most of them have that propensity. But if, for example, it has a very nice way of breaking up the database and the redo log, you put it in a different place and use different technology. Because it's sequential workload, exactly. You wouldn't want to put that necessarily in flash. Yes. And so it's very smart about how it does things. So it's a more traditional tiering array, but the performance of the flash components of it are very good indeed. So it really is purpose built for Oracle environments. Yeah. Okay, so that's good. So let's take a look at Oracle storage today versus say, when did Oracle acquire Sun 2010? You want to say? Was that right? And they acquired... They acquired... Or was it... Pillar in 2011, yeah. Yeah, okay, so at the time, Sun storage was like, eh, you know, I used to talk to McNeely all the time, you got to get your act together in storage. It's like, well, I have a store or company. Take some time. You worked at IBM, you know, the challenges. But really the lineup was pretty mediocre. Now with the investments that Oracle has made, first in the NAS side with the ZFS appliance, I'm bad with names, Z-E-S-3-2 or something. I think that's what they're called. That is a flash first array. Yep. Good design, some really strong use cases for backup, particularly... Good bandwidth. Very high bandwidth. Analytics. Yeah. But the big gap was on the block side. That's what FS1 fills. So I have said earlier on theCUBE, a lot of customers that I talk to, you ask them what their storage strategy is and they say, EMC for block? NetApp for file. That's the strategy. It's like you got FS1 for block and ZFS for file. That's exactly right. Is that kind of the strategy that Oracle customers are going to take? Absolutely. There is a capability within this block of doing file as well. But that is not their focus and it shouldn't be their focus. They've got two good lines. They're sharing a lot of the technology, the disk drives, et cetera, and the disk enclosures. They're sharing that between the two of them. But there's two separate products, both of which do different jobs for Oracle applications. So Matt, can we call up that slide that we've been showing intermittently today? So this is David, just so you can see. Yeah, I remember that slide. Larry brought it up Sunday night and Mike Workman brought it again up yesterday at his discussion at launch. We've got the Extreme I.O. on the left side, the Oracle on the right side. These are Oracle benchmarks. They're renowned for this, basically saying we're 10 times faster, it's five times better, it's half the price, and that's not all. So Mike Workman was on here and Stu and I were trying to get it word and edgewise over his enthusiasm for this because he's obviously very excited. It's like giving birth. It's a good product. It's a great product. I said, floor is coming on later. That's nice that you say it's half the price. So I want to get your analysis on, I mean, everybody takes this stuff with a grain of salt. Although, I bet you there's a lot of DBAs in the audience that are nodding their head and going, okay. So what's the, I don't know if you've had time to peel this back, understand what the workload is underneath. I don't know if precisely what work they're doing, but first of all, there are two types of flash in the box. There's a capacity flash and there's the performance flash. So if you're going for price, if you're going for price, you're going to go for the capacity flash. And if by chance the workload was an HCC type workload, which is mainly read, that would be a very good use of the capacity flash on that particular box. So horses for courses, that would be a very good fit indeed. That would bring the price down very significantly. Okay, but let's dig in. Is this sort of a marketing chart? Is it still up there? I can see it. So it starts with capacity. There's a two node building block. So you've got capacity, read, IOPS, write IOPS, read, write, 50, 50, read, write, 50, 50, all IOPS, and then gigabytes per second. So bandwidth, writes, and reads. So you got IOPS, you got bandwidth, there's no latency. Mike Workham says 500 milliseconds. Microseconds, sorry, latency. Yeah, sounds very reasonable. I'm not sure why they didn't show any latency figures. No, again, if it happens to be that workload, then you'll have very big blocks, and you'll be doing them as big blocks and the IO time will be a lot longer. So these are 32K blocks sizes. Blocks, which are the normal database blocks size. Although he pointed out that redo logs, for example, will be bigger block sizes. Much bigger blocks. And I suspect for this one, it'll be much bigger blocks as well. I'm going to put words on you both. You're saying this is rigged in favor of the workload. You know, they're a good marketing side, and it'll be truthful that at this particular moment, using this particular workload and this particular features of their product. But if this, if this, and this, and this, and this, if this, then we smoke on. Yeah, we smoke on. And it's always horses for courses. So okay, so knowing what you know about the FS1, knowing what you know about the Extreme IO, what are the horses for courses? Talk to customers. Well, so if you're going to take a online application, a LTP type application, in that sort of area, deduplication doesn't give you much. So it's going to be very good. Compression is actually the better of the two. So both neither of them at the moment, the Extreme IO or the I'll have compression from the point of view of the piece of the workload you want to do for your analytic stuff that will have compression, that will go very fast. So I'm alluding you, I'm alluding you. Let's back up. You got Oracle FS1, no compression, but they use Oracle Hybrid Column compression. For some of the workload types. By the way, less than half the price, I'm presuming they Oracle had compression on. If you have the right application, you can get 10 to one. Do you know what was assumed for compression in these? No, I don't. But I would suspect, I strongly suspect. You strongly suspect Oracle had compression on and Extreme IO had it off and because it's not shipping yet. If you have compression on, then your bandwidth goes up and your throughput goes up and your cost goes down. So what can people expect? I know there's no typical, but what can people expect on average for compression rates? Compression rates for 3G1, 3G1, for an OLTP type. Yeah, okay, and whether it's hybrid columnar or done in the array, you're going to have similar type of? Well, the hybrid columnar is much bigger than that because it can only be done on analytics and workloads like that. Ah, so not OLTP. So OLTP is row-based columnar. Yeah, so what's the Oracle strategy for compression in an OLTP environment? Well, there you use much smaller blocks. Your compression rate goes down. What am I, what's my? You use two to one to three to one. So how are they doing that? Is that in the database still or? No, oh, well, no. That would, I mean, I don't know how they're going to do, they haven't got it for row. Okay, so that would be an advantage for a product that has compression? Has compression, like Pure, for example. And probably Extreme IO is coming soon. Isn't everybody going to have this? Absolutely. It's going to be like table stakes, right? You wrote a piece on the other day on permabits appliance, data reduction appliance. I mean, so anybody really wants it, can go apply it. In my opinion, Oracle should put a permabit type solution in front of it, if they don't want to do it themselves. It's a very good solution, and for some workloads, it'll smoke other things, yeah. Okay, so you need more time to peel back. Exactly what's going on. But my experience is it's a good product. It fits Oracle extremely well. It's a nice fit in terms of, and the ability to have multiple tiers for Flash is excellent. Okay, but now, so I'm an IT practitioner, I'm a guy in the Wikibon community. I call you up, hey, David, I read your piece in the FS1. I saw I was at Oracle Open World at Trash and EMC. When should I use each? When should I consider FS1? When should I consider Extreme IO? So FS1 is only high performance Flash, obviously. So. FS1? Sorry, sorry. Extreme IO, yes. So if you've got certain workloads where you need the lowest latency possible OLTP, then probably the Extreme IO is going to do a better job in that one area. It's not going to be a huge difference between the two, but where latency is critical, then Extreme IO is probably going to do very well. If you've got a lot of different workloads and you're spending a lot of time setting them up, tearing them down, then the automation that you get in the FS1 is excellent. It really is. I mean, one thing. It's particularly within the red stack. Absolutely. Now, what about things like snapshots? You mentioned that. Snapshots, they have a set of snapshot capabilities. Well, Extreme IO have very good snapshots. They have the very space-efficient snapshots and one of the things that they're emphasizing is the ability to share the same Flash data over multiple applications. So you can share the OLTP with the analytic, with other things. So that's an area where their type of snapshot does better. Oracle, I'm sure, are going to have the same type of technique. You have to with Flash, but it's more of a cloning type. Do you know if it's in there today or you don't know? It's more of a cloning type, the... So it's not as space-efficient today? It's not as space-efficient. Okay. So in that area, but again, the two are going to collide. They're going to have very similar features. So you didn't mention sort of what you typically talk about with Oracle versus EMC or others like NetApp, the horizontal support for applications across the portfolio versus the narrow vertical. It sounds like Oracle FS1 is tuned for the narrow vertical red stack, but Workman said on theCUBE today, oh, we support VMware. We support Microsoft, Type-RV, we support other environments. That's strategically, we're going after that. So... In order to support those, you're going to have need, deduplication, and compression. Outside of the database. Outside of the database. So what do you make of? Take the hybrid columnar compression, for example. What do you make of the fact that Oracle basically locks out its competitors from taking advantage of hybrid columnar compression? Before Oracle owned Sun, of course it never would have done that because it would have wanted all the array companies to have hybrid columnar compression and take advantage of its software. Now that it owned Sun, it plays that card. You think that's a good strategy or is it dangerous? I think it's a dumb strategy. Why? Because they make far more money from the database. Then they do from the storage. And pushing the database and getting them into the data warehouse and analytics stuff, that's under real pressure from a whole lot of other players. So why not have it working on all storage and improve the productivity? People are not using that. That would be a benefit to customers. Oracle is always talking about how Larry said, ah, we care about our customers, we care about our customers. If you really cared about your customers more than gaining storage market share, wouldn't you open up hybrid columnar compression to everybody? Every day of the week, yeah. So I mean, it's clear. Oracle wants to gain share and storage. Okay, but eventually there could be pressure. That sort of forces the other storage vendors toward Microsoft's camp. Microsoft said, hey, we'll let you take advantage of our bell and whistle, whatever it is. So to the extent that the Microsoft database becomes more competitive from an ecosystem standpoint. Oracle database is still a damper database. Okay, then if Oracle still got that much of a lead over Microsoft. In the high end. Okay, so DB2 is still narrowly focused on IBM customers. Then why not use it as a lever to gain storage market share? DB2 in this very area of hybrid columnar compression is actually doing extremely well. They've got a really good announcement with DB2 for data warehousing application. So. The player. Yeah. No doubt. Okay, so that's good. We covered a lot of ground there. What about NetApp? You know, NetApp got a big booth here. No, no, no EMC booth at the show. They decided not to invest this year. Maybe they'll be back next year. EMC has put a big effort into Oracle in the last three years. Probably decided to take a little breather, save some cash. Who knows? But NetApp's got a big presence here. What do you see from NetApp these days? Well, NetApp has, as you said earlier on, NetApp and the file, they are very, very deeply entrenched. And a lot of backup is done on NetApp boxes within a huge amount of time. And Oracle's going after NetApp. You know, we had an Oracle executive on at VMware, VMworld, and Oracle. Wants to share. Used to be one of NetApp's biggest customers. NetApp was a reference customer. NetApp talked about them a lot. Dave Hitz, for years, talked about Oracle. So they had an Oracle affinity. And the guest in the queue said that Oracle IT swept the shop of NetApp, brought in the ZFS array. Exactly. Whichever one it was. And he gave, you know, all kinds of sort of metrics on how much money they saved. He said, now, of course, you got to be, again, those metrics are yesterday's NetApp with our today. But still. It's the key for backup. But it's right. And so NetApp lost a big account. And it looks like Oracle for the NAS stuff is going hard after NetApp. It's one of the areas that's growing, according to SAFRA cats in the conference calls. Not much within hardware was growing recently, but that is super cluster, I think, is. Obviously engineered systems. And we'll see about FS1. But where is NetApp these days? What do you, I know you spent some time. On the database side, which is the interesting one. Yeah, so yeah, let's talk, actually, let's talk NetApp specific within Oracle. And then generally with their NetApp. Well, on the Oracle, we did some research recently on the leadership, Oracle leadership. And we asked people with Oracle, heavy Oracle users, where Oracle was an important component. We asked them to rate who their best supplier was and a storage supplier in infrastructure. And on the storage side, EMC got a very high percentage and Oracle was came up very strongly as well. Really? Very strong, both of them. 42% for EMC, 16% for Oracle. NetApp were way down at 5%. So that shows you for the high end Oracle, NetApp are not really playing. Now, in the lower end, of course, they have a very good story. But in the high end, they're just not playing at all. Similarly, when you're looking at the server, what they picked out as the best server, Cisco has done extremely well. Whereas before you'd accept IBM to be up there in HP, now it's Cisco who's competing. But when you looked at their choice of VCE versus, for example, the NetApp combination with Cisco, it was a eight to one difference between the two. So again, in the high end, they're having a challenge. It's in the lower end. It's in the NAS style that they're doing extremely well. It's interesting, I'm looking at this chart. I just put it out on CrowdChat. CrowdChat.net slash OOW14 is the hashtag. This chart about the single most important factor for Oracle infrastructure leadership, reliability. Reliability is everything off the top. This is for Oracle intense environments. I mean, everything else pales in compare availability, high end performance, virtualization. Same survey that we did the other questions on. Yeah, you googled it and it came up here, but reliability for infrastructure, that's interesting. For Oracle, that's it. I mean, that's why they sell the rack. It's so reliable. It's software reliability that is most important for those guys. And that's another reason why the FS1 works so well because it works well with rack and it works well with all of the software. It doesn't have to do the sort of work that EMC does. It avoids it. So you have this other chart in here on MindShare. MindShare, that's the one. You see EMC is up at 42% and Oracle at 16. So what are you measuring here? What is this? This is the MindShare. You're single. So you said who leads in infrastructure? Yeah, who is the, you think does the best job in Oracle infrastructure and storage? And there's a MindShare one and the same in one of the papers that I did on the servers as well, which shows VCE coming up strongly. That's interesting. So it's interesting, a couple of things. One is that strategically, Oracle's going after NetApp. I guess it makes sense. They get the NAS product. They go after NetApp, but this FS1 clearly positioned to go after EMC. EMC's number one in Oracle. Number one, clearly. Yeah, storage company, by far. Yes, by far and away. Yes, and the more Oracle you have, the higher the availability and reliability, the more EMC you have. So what's your prediction? Well, right now EMC's number one in block and NetApp's probably number one in file. Yeah. Will Oracle be number one in either of those within the next five years? No? No, why not? Because there's so many other workloads out there other than Oracle, and they're growing faster than the main frame. Where they'll say, okay, we got this other main workloads, we'll put Oracle on there too. Yes, yes. So I mean, if you look at it as Oracle as being the main frame type of software, all the other software is coming around it and it's growing much faster. Okay, so that's one question. Next question is, will Oracle gain meaningful share in storage in the next five years? Yes, absolutely. In its space, I think by combining the software and the hardware together and making these stacks, the exodators and the FS1 will go into that same stack, it can share things more easily. Who gets hurt? EMC, NetApp or others? EMC. You think EMC will lose share within the Oracle? It will lose share in the Oracle space. Really? Very significant. I mean, it's not common. It's so believing. It's not common that EMC loses share in anything. They always seem like they get booted out of Dell, they go after Dell, they get knocked out of the OEM relationship with HP, they go dominate HP. They have a knack of doing that. You think it's more challenging to do that? What they're doing very well is the VC and they will use that as a general purpose way of getting into that area. But their proportion of, if you look at EMC at Cisco is a lot high percentage of that. I mean, we haven't even talked about these other things. We're out of time. Can I go a little bit longer? I'd like to, I got to get your take on Oracle Cloud. Oracle, Sunday night Larry Ellison, I don't know if you heard his talk, said basically, we had to be, we made a promise to our customers to be in all three layers of the cloud, infrastructure as a service, platform as a service and SAS. They're kind of defining platform as a service database, Java, other tools. Yep, very sensibly. That's a good proportion. So you look at that as pass. I mean, you see what like what cloud foundry is doing. You see what Salesforce is doing, Oracle's right there with them in your view. I mean, scale, web scale. It's providing a slightly different service, but they can provide the software and the software maintenance and the application, their applications. They can provide a soup to nuts single throat to choke and it's very complex. And that's something that nobody else can do. So they're working. So let's talk about the infrastructure as a service. The Oracle said that they will price consistent with Amazon and Google and Microsoft. They don't have the volumes that those guys have, not even close. I had heard that they have like 20 or 30,000 servers in their data center. I don't know how many Google has, but it's more than 20,000. It's got to be hundreds of thousands, right? Yeah, yeah. Maybe they're pushing a million, but so they... But look at where the cost is for Oracle. Look at the maintenance costs, for example, for them to maintain their own products and all those different platforms. If they can make this platform, the priority platform, they can cut down the amount that they spend on maintenance on their products by a factor of two or three. That is a huge saving to them. So I feel like, well, we know that one of the advantages that certainly Google and Amazon have is the homogeneity of their infrastructure. Probably to a large degree, getting to be the case at Microsoft. But here it's the homogeneity of software and the services around it that is going to be the killer price tag. Can IBM replicate that homogeneity? If it had Oracle it might, but it lost out to... If it had the Oracle database. It's the software here that is the... If you look at the cost of the infrastructure for an Oracle application, 70% of it is the database. Yeah, but IBM's not paying that cost for its own database, right? It's using... Right, they have to get them to convert to DB2. Well, but IBM has its own cloud and IBM... Sure. IBM can offer databases to service with DB2. But the price of the profitability of that database. Right, so you're saying when IBM offers Oracle as a service. It has to go to Oracle. The customer's paying Oracle the money. Or IBM is paying Oracle the money. So Microsoft's got the advantage there, obviously. Absolutely. People are using SQL Server. Okay, so... And it's got its ordinary software of exchange and all the things like that. In infrastructure for service, Oracle, from a volume standpoint, can't compete against Amazon and Google. But it owns its own hardware. It owns its own hardware. And it owns its own software. And it's the software costs that it can bring down. So it can... So economically, it should be able to compete. I believe so, yes. Because it's got software margin of economics. It owns the software Amazon. Customers are paying Oracle for Oracle. And Microsoft owns so much of the software as well. I don't know. We're going to reinvent in November. We're going to have to poke at this a little bit. I know Amazon's pretty forceful about getting software companies to list in the marketplace. Many do. I'm not sure Oracle does. I kind of doubt Oracle does. It's going to bring your own... They have to bring your own software. I think they designed to bring your own software for Oracle. They have an agreement to have a software, a sort of Oracle by the car. Oracle by the drink. By the drink, yes, within it. But Oracle, of course, can compete very effectively with that for a larger drink. For a pint of beer as opposed to a sip of beer, yeah. Yeah, I got to understand how that works. Because when you look at, again, the economics of this, the whole cloud thing, I think of VMware building out its own data centers for hybrid cloud service, VMware slash EMC. I look at the collection of 4,000 service providers, all smaller than Amazon, and saying, it's going to be tough to get 4,000 companies, all homogeneous on VMware, when many of them want to use Zen, OpenStack. Well, take an example, for example, for Oracle, the ODA. You know, the Oracle DataPlace appliance. They can reduce the price of their software by virtualizing the cores. And then they can say, you only need two, you can only have two. And on top of that, the ISV can reduce its prices in terms of being able to reduce the amount of effort required to do that. So the net effect of that is that it's the higher up the stack you go, the more you can cut of that price. What I've written about is SMEs, single-branded. Yeah, you've said there's more value the higher up the stack you go. So as long as you have the applications and the database, and you can get the ISVs to come in and use your service, you're going to be in a very good position. So there's a big chunk of market that they can go after and do extremely well. So you think that, don't be off put. My daughter just texted me. My daughter who's now at college, and she asked me how I was, so I was sending her pictures of the cube, live in the cube. So I don't get to talk to my daughter as much anymore, now that she's off in school. But so you think that they can effectively compete economically because they own the full stack. That's right, yes. And Microsoft can as well. That doesn't mean to say there won't be new players in this field for the software, because the ISVs can now go to much cheaper places like Amazon and develop new applications with new functionality. That's a whole different story. It's interesting, you know, I mean, year ago Workday was looking like awesome. Oracle's now sort of, she'd be marketing, positioning, and doing so very effectively now. I mean, we're a PeopleSoft user. We know what it's like, it's kind of clunky, not the best interface in the world. But it was great in the client server day, but so I don't doubt that there's some other advantages out there. We were at Infor a couple of weeks ago seeing some of their beautiful, elegant software. By vertical, by micro-vertical. But Oracle, you know, they got these big companies, they got cash, they got marketing, and they got customers. Yep, and if they take that strategy and go down into the micro-vertical with that amount of cash behind them. A lot of observers don't like this talk track, but it's hard to, it's very hard to predict like there's going to be some total innovators, dilemma, disruption, because of cloud to Oracle. They've jumped in just like Microsoft has. We're getting the high sign, and we're getting the hook here, David. All right, okay, good to talk to you. Great to be touching a couple of topics anyway. Thanks for your insights. And always a pleasure. Check out wikibond.org. David Floyd's got a bunch of research up there. Oracle leadership, stuff that he's done, the FS1 analysis. We've done tons of stuff on Extreme IO if you want to compare sort of independent perspective. And so again, thanks for coming on. No problem. All right, keep it right there, everybody. We will be right back after this word. This is theCUBE. We're live from Moscone. This is Oracle Open World. This is theCUBE.