 TheCube at EMC World 2014 is brought to you by EMC. Redefine VCE, innovating the world's first converged infrastructure solution for private cloud computing. Brocade, say goodbye to the status quo and hello to brocade. Welcome back to Las Vegas everybody. This is Dave Vellante, I'm with Wikibon.org and Jeff Frick is my co-host for this segment. Don Sullivan is here, he's a staff systems engineer, database specialist at VMware. Don, welcome to theCUBE. Thanks Dave. So we're going to talk about what's new with VMware and Oracle. Many of you might remember several years ago now, it had to be at least three years ago. We started to investigate the merits of virtualizing Oracle and at the time Oracle was very negative on virtualizing, well, virtualizing Oracle database based apps with VMware. And so we looked at that, we talked to a number of customers who had successfully done it. There was a lot of fun in the marketplace that Oracle wouldn't support it and of course that's all proven not to be true. And so we wrote a piece called Damn the Torpedoes, virtualized Oracle ASAP. And since then the uptake has been amazing, it wasn't our piece necessarily did it but I think we called it right. And the business value that has been created has been tremendous. So Don, with that as introduction, does that align with what you recall a few years ago and the uptake now of VMware and Oracle shops? Oh, absolutely and being very humble, Dave and that your piece did actually was actually a major aspect of how the adoption rate has been increasing tremendously. And in fact, I would refer to it this way, the adoption rate is increasing precipitously as well as the increase of the adoption rate is increasing amazingly. I've spent some time at a number of sessions at the Collaborate IOUG conference a few weeks ago and the number of hands raised when you asked the question as to how many of you folks in a room, Oracle folks, are actually virtualizing and then virtualizing with vSphere 50, 75% and we're seeing that here in the Oracle sessions as well. Yeah, so talk a little bit about that journey, we see worlds, so we like to talk about journeys. What's the journey been for Oracle shops? Where did they start and now you're saying that the pace of adoption is actually accelerating. So where are they starting and where is it headed? Well, it's all kind of started I'd say back in the 2008 timeframe, certainly with the release of vSphere 4 and the capabilities that were resident in vSphere 4 to really minimize the overhead and the impact of the virtualization layer on the overall infrastructure, really made it at a point and as time went on with various different customers and internal tests and we started recognizing that and the customer base started recognizing it, maybe there was a two, four, six, up to 10% overhead and depending on the metric and the various dimension of performance you were looking at, it was very, very minimal and as the releases went on, four, four, one, five, five, five, it really stopped becoming a discussion about the performance capabilities and sort of moving into the discussion about the value. In fact, one of the books that I'm writing here actually talks specifically about the value. We own the letter V since we're VMware and we talk about the value and it really falls into two categories, the resource management capabilities that exist in the infrastructure and then the various different features that allow you to meet your SLAs, whether they be availability, disaster recovery, performance, security, or any other SLA that you want to conceive of and that's where we're at right now. We're at a point where 99.9% of every database management system on Earth makes sense to consider putting in a virtualized infrastructure. Yeah, so if we go back to the sort of original research we did what we found that was fascinating. So Oracle of course put a lot of thought out in the market. We put fun at Oracle a lot but the FUD was we're not going to support that. If you have a problem, we're going to force you to go back to a physical environment. I don't know of a situation where that's ever happened. Do you? It's not impossible it could happen but it's extremely rare and it is FUD and it's fun in every possible level in the sense that if you had a situation in which say for instance a legacy application, Oracle has somewhere in 100 plus applications that they've acquired at some point. If you had a legacy application that say I had a specific support statement that it only supported specific types of infrastructure you might actually run into that. I've actually run into it once to be honest back in 2009 but it was no merit to it whatsoever and in fact after our, meaning VMware's, GSS Oracle support team which by the way is the best support team I've ever worked with and I spent seven years at Oracle, five years at HP and so on. The GSS Oracle support team which is an Oracle support team again is just amazing. Once they get involved with it and they will open a ticket and they will hold the ticket open until it's completion under all circumstances, no matter what. Once they get involved with that within a couple of days that particular environment with that legacy application well Oracle realized it was a problem that was in Java code and they just took it back over. It's not an issue. This is the point I wanted to make is that while Oracle sort of, and it was mostly, a lot of it was the sales guys. Always the sales, yeah. But the reality is the customers that we've talked to to your point of Oracle support has been phenomenal. And Oracle is a great company. They're very customer driven and they understand the complexities of database based applications and the importance of not losing data, protecting data. And Oracle has a pretty robust stack in that regard. So our experience has been that the support experience from our customers has been quite astounding. Absolutely. There just is, it's a red herring often that comes through when it gets confused but there just really is not an issue with it and it never has been. Now you had mentioned off camera that you got kind of working on some new reference architectures with Cisco. Why don't you talk about that a little bit? So between ourselves, VMware, Cisco, EMC, run by a third party testing company called Principal Technologies out of North Carolina. We've done a number of different reference architectures, functional stress tests is what we refer to them. The most impressive of which is the functional stress test which we took of VNACs and Cisco B2M300 blades using of course all the Cisco networking and we were in RAC, it was 11203 at the time, we started this last year and we used a background tool, a workload tool called Benchmark Factory. And we took three RAC nodes which of course were virtualized so they were logical RAC nodes and we put them under as much stress as possible. And what we decided to do is let's see what happens when we do a VMotion. So anybody who knows RAC knows this is extremely sensitive towards network disturbance of some kind, it's purposefully built that way. So if the disturbance is gonna show up and if anything's gonna happen, it's gonna happen during a VMotion operation. So we did the VMotion under extreme stress from basically taking a node on physical host A and moving it to B. Nothing happened, everything was fine. Then we went a little crazy and decided, well, we're gonna do two of them simultaneously. We're gonna go A to B and then B to A. Nothing happened, everything was great. And then we went really crazy. And we said, here's what we're gonna do. We're gonna send all three of them together simultaneously concurrently and we're gonna keep them going in a round robin fashion. A to B, B to C, C to A under extremely high stress and what happened? Nothing. Throwing the complete 100% abstraction between the operating system and the hardware even for something as sensitive as Oracle RAC even under the most stress that we possibly could throw at at the same time. So really showing a lot of things you can infer from that, specifically the reality of with vSphere, your hardware downtime is pretty much eliminated. And other things as well. Oracle RAC, real application clusters is, as you say, probably the most demanding Oracle environment. And talk about why it's so challenging. It's all about recovery, right? When something goes wrong, how do you recover, right? Well, and in this case, fortunately we didn't need to do any recovery because nothing ever happened. So typically under the circumstances of network disruption with Oracle RAC, which is again, like you said, the most sophisticated, powerful database management technology in human history. And if something would happen, a network disruption of some kind, you get the individual RAC nodes to eject from the cluster and be fenced from the storage and they have to reboot and they have to be recoverable. Even under those circumstances, we didn't necessarily need this in a test because nothing happened to cause a recovery situation. But even under those circumstances, that Oracle RAC node would just simply, the VM actually under it, would just simply be restarted on another available node in the ESX cluster, just like any other node. And it would be back up and running in four, five minutes with completely seamlessly and you'd be back up to full strength. So Don, I want to ask you, so VM were obviously great penetration into Oracle environments. We were very excited about that because we felt like it was the right thing for clients. You obviously must be excited. But now you have Oracle coming in. It was funny, you know, Larry Ellison at the Churchill Club with Eddie Zander ripping on cloud. I'm sure you saw that, you know. And now you've got a total 180 by Oracle there, going hard after cloud. You got the red cloud. How do you see that affecting VMware adoption, if at all? I don't see it at all, particularly when it comes to the 12C features. And I like to, you know, I worked at Oracle. I was in the structure at Oracle for seven years. And we kind of joke now that Oracle after 37 years has finally recognized a relational model. And they've got the idea of the pluggable databases and the cloud aspect of that and the ability to move databases as individual entities around. None of that changes the overall value proposition of virtualization. In fact, I would say that there's every feature. Rack, the pluggable databases or anything else that you can conceive of is completely complimentary with virtualized infrastructure. Especially when you're dealing with what is here a non-paravirtualized type one hypervisor like ESX, which totally again abstracts the operating system from the hardware. So I think you referenced before a session here on Oracle and the adoption. When you go to the independent Oracle user group meetings, what are you hearing from customers around VMware and Oracle? Well, this goes back to the original question as well. Three, four, five years ago, well, there was a lot of resistance. There was a lot of dismissal. I did the very first real presentation where Oracle extended the existing support statement to Oracle Rack in 2010 for 11202 and above. So after that point, we started doing a lot of public discussion about this. And I did some of the original presentations of running Oracle Rack on vSphere and talking about how complimentary they all were. And there was a lot of resistance. It was almost emotional at times. That resistance is almost gone now. Three years ago even, I remember we did six separate sessions at Collaborate and IOUG in here. I believe it was in Phoenician in which we had maybe 50 to 75 people in the audience each and they were all new to this. None of them really believed it. The typical Oracle DBAs being very, very conservative, right? I refer to them as the Praetorian Guard of Data. They're so conservative in protecting. But now, again, a few weeks ago at the Collaborate conference here as well, it was just amazing. The rooms were packed and it was about 75% of people who at least either were considering it or had already done that virtualization. And honestly, I wish at VMWare, we were a little better at customer collection of data in that sense. I really want to know the actual rates of adoption and the actual total number of the customer base, but it's so hard to sell many. Well, you sell through the channel exclusively, right? So it's hard to track that stuff. Well, we're going to be doing some surveys at IOUG to actually try to capture some of this data soon. Well, it's an awesome user organization. Some of the most advanced customers in the world. Okay, Don, last question is, put a bumper sticker on EMC World 2014. The truck's pulling away from the Las Vegas Convention Center, the Hans Sands Convention Center here. What's the bumper sticker say on 2014? Flash into the future. Ha ha, good, good stuff. Yeah, we didn't talk about flash. No, but there's only going to flash, okay. Can we have a quick discussion on that? Yeah, we got, we okay? Yeah, talk about flash. What, I mean, if you think about flash and the impact on database, generally in specifically Oracle environments, you got to be excited about that. So flash done right, effectively eliminate storage configuration and the age old problems that database administrators had with slogging their way through how the storage is set up can completely go away because not only of the simplicity of it, but obviously of the speed. And it simply makes all of everything we talked about in the last 15 minutes a better overall proposition. In fact, the next of these studies, I talked about the Rockin vSphere study, the Mega V motion study, the next of them we're doing in the series is going to be a mass provisioning study with the Cloud Automation Center and the same infrastructure I mentioned, except we're going to be switching out the Vmax for an extreme IO array and showing how simple the entire logical infrastructure is with effectively the databases acting as props. This will be SQL server as well as Oracle involved and then on the stream IO array. So this is where there's a huge opportunity I think for clients is, and we've written about this on Wikibon. If you bring in flash and actually beef up your storage infrastructure using flash, whether you're putting it into an existing array or frankly ideally going with an all flash array, maybe spending a little bit more on your storage infrastructure, what we found is that you can reduce the number of cores required to support your installation. Now how does Oracle price, it's database licenses, it's by core. So if you can reduce your number of cores, you can reduce the amount you're paying Oracle in terms of database licenses and of course that'll ripple through maintenance. Now why is that important? It's important because when you talk to customers, many customers 50% of their TCO, not their CAPEX, but their TCO is Oracle license and maintenance. And so if you can use flash, maybe spend a little bit more on your storage infrastructure, optimize and balance out that infrastructure, reduce your cores, you're going to have a direct positive impact on your license and maintenance fees. Have you seen that? I absolutely have. In fact, there's something that is absolutely both implied and can be inferred if not explicitly stated in your first work, way before really the incredible adoption of flash right now. So that was a tremendous foresight on your part. But also, I mean, I'm seeing this as not just an initial interest point of customers but one in which they're just clamoring for more information about. So the very paradigm of virtualization is to make your infrastructure more effective. But when the infrastructure itself becomes more effective at the same time, everybody wins. Yeah, and by the way, that's not exclusive to Oracle database environments. It's true for SQL server, DB2, et cetera. So, all right, Don, listen, thanks very much for coming to theCUBE. It's a pleasure having you. Absolutely, thank you. All right, keep it right there everybody. We'll be right back with our next guest. This is theCUBE and we're live from VM, VMworld, EMCworlds, 2014. Yeah, VMworlds in a few months. All right, keep it right there, we'll be right back. Okay.