 Live from San Francisco, California, it's theCUBE at VMworld 2014. Brought to you by VMware, Cisco, EMC, HP, and Nutanix. Now here are your hosts, John Furrier and Dave Vellante. Okay, welcome back everyone live here in San Francisco, California. This is theCUBE VMworld 2014, our 50-year broadcasting live from VMworld. I'm John Furrier, my co-host Dave Vellante, our next guest, Steve Zavanec, Steve Zavanec, VP of Storage, Hardware Business Group at Oracle, welcome to theCUBE. Always great to have Oracle on theCUBE. Thank you, John, pleasure to be here. I got to ask you a first question. Has Oracle been on theCUBE or is this a first? That is a first. Isn't it? This is a first, Oracle's on theCUBE. I think so, congratulations. It's not your first, you've been on before. I mean Oracle, I mean Larry Ellison definitely would be great theater on theCUBE. He would be fantastic. Certainly would hold his own. I think he can hang with us. John, do you think you can go toe-to-toe with Larry? Yeah, absolutely I can. No doubt I can go toe-to-toe with Larry Ellison. Benioff can't. There's no problem, any tech athlete, Dave, you and I can handle on theCUBE. So Steve, first question for you, Oracle. I got to ask on a technology front, virtualization environment, VMware is the show. What's the deal? How do you guys, how do you guys interplay on the virtualized world of VMworld? Sure, so if you look at VMware, if you look at a highly virtualized environment, Oracle right now has the only stored system on the floor that's an in-memory stored system. If you look at what we bring to a highly virtualized environment, we're booting 16,000 VMs directly out of DRAM, directly out of memory. And we're doing that in under seven minutes at a list price for less than 200,000. Also, we can boot 20,000 VMs under 10 minutes at a list price of, again, under 200,000. You contrast to anybody else in the industry. They're not using a memory-centric architecture, a DRAM-centric architecture. We are the only storage in memory-centric architecture on the floor right now. And if you look at, let's contrast that to NetApp. NetApp boots about 1,000 VMs in 20 minutes. We're booting 16,000 VMs in less than seven. So the fact that we're driving all of those boots, the VDI boots directly out of memory, is a significant benefit for companies running highly virtualized environments because it accelerates their time to deployment. It accelerates their time to getting business up and running each day if you wake up and you can boot tens of thousands of VDIs in mere minutes. Just run those numbers by me again compared to NetApp. You're booting up from in-memory. We're booting in-memory. How many VMwares, VMs? 16,000 in seven minutes. Seconds? Seven minutes. Okay, got it, okay. NetApp's doing 1,000 approximately in 20 minutes. So 16,000, seven minutes. NetApp's doing 1,000 in 20 minutes. So that's the advantages in a highly virtualized environment with a memory-centric architecture. So I was walking the floor the other day, I saw the Oracle, you guys got a big presence here. I was like, pinch me, Oracle of VMworld. So a lot of your customers run VMware. And early on, there was some nervousness about it. Many customers, we advised them, Wikibon, look, virtualize Oracle databases, it's a good experience. Oracle will support them. Consistently, customers have said, yeah, Oracle's a great job supporting them. But what do you bring specifically to the table in a VMware environment from a storage perspective? So if you look at it, Oracle has extensive experience in our own highly virtualized managed cloud services environment and our IT production environments which are highly virtualized in terms of what we've been able to do with virtualization and our storage. So if you think about it, we were running NetApp before. We had about 50 petabytes of NetApp about three years ago. Inside your own shop. Inside Oracle, worldwide. You're a reference customer. We were a reference customer. And so that has, we've consolidated all of that NetApp, moved it, shifted it all to Oracle's EFS storage appliances. So from a VM perspective, we achieved a nine to one VM consolidation ratio. We were getting about 250 VMs per NetApp Fast 6000, at which point the CPU would be pinned at 100%. And then we'd have to buy another NetApp filer to store another 250 VMs. CPU would get pinned at 100%. And another filer, another filer, and that's how you get filer sprawl. With the ZFS storage appliance, and this is the prior generation, not the current generation ZS3, we were able to achieve 2,300 VMs per individual system. While the CPU utilization rate was only 33%. So if you think about it from a VM density standpoint, again, physical systems, three to one consolidation ratio versus NetApp Fast 6000 and the ZFS storage appliance. VM consolidation ratio, we went from nine to one, 2,300 VMs to, rather going from 250 VMs to 2,300 on our ZFS. So what we bring to the table is extremes of VM density and the ability to boot VMs directly, directly out of memory. That's a distinct advantage. If you look at our exadata, it's leveraging in-memory architecture, the layer of flash. If you look at our ZFS, it's leveraging in-memory architecture with a layer of flash and HDDs. You look at our Spark M6 big memory machine. It's 32 terabytes of memory. So Oracle is significantly leveraging memory and memory computing across its entire portfolio. And the combination with tiers of flash and HDD brings significant benefits to these types of highly virtualized environments. Okay, so you're talking about, you've mentioned NetApp several times, so you're talking about file storage, you're talking about NAS, Brock. So which product are you talking about specifically in your portfolio? What are you featuring? We're talking about two models. We have the ZFS, ZS3 series appliances, storage appliances. That's the ZS3-2 model, our mid-range model, and then our higher end models is ZS3-4. And we compete with, like I said, the likes of NetApp's Fast 6000, which we have superior performance, efficiency, VM density, and the likes of EMC's VNX series, for example. And we're just talking about the VMware differentiators. A lot of customers, obviously, as you said, are running Oracle Database in VMware environments, right? And if you look at Oracle Database, here's an example. If you think about analytics, okay? Oracle Database, it writes to storage, and if it doesn't write to storage within 200 milliseconds, it times out. Our closest competitor has an analytical tool that checks for data, scans for data every 30 seconds, and NetApp has one that takes five minutes. So if you time out every 200 milliseconds, right, where you're trying to write to storage, that means within that 200 millisecond window, there are 150 chances of timing out, 150 times of timing out within that 30-second window, if you think about it. So if you've got 30 seconds for your product to run analytics, within that timeframe, there's 150 chances because of the 200 millisecond Oracle timeout in terms of writing to the storage. Now, the point here is, if our closest competitor takes 30 seconds for their analytics tool to do a scan on the storage environment, and ours, our detrace analytics software, which we share in conjunction with our Spark servers, that only takes one second. So from a VMware perspective, we're integrated with the vSphere console, and we're scanning every one second. So directly from the vSphere console, people can deploy our detrace analytics suite, software analytics suite, and immediately find out what's going on at the VMDK level, the file level, the NIC level. And so the ability to work at those extreme velocities in terms of the high-speed analytics that we have that come with the ZFS, makes a huge difference in a highly virtualized environment. If you think about it again, 200 millisecond timeout, that means in a 30-second window, you got 150 opportunities for timing out when you haven't achieved and obtained any shred of data that you can analyze and find out what's wrong with your environment. So you obviously, you know, Target, you said your closest competitor, you're targeting NetApp obviously for the ZFS because it's file storage. And the ZFS is a flash-first architecture. Meaning, when I say flash-first, I mean, the ultimate right goes to the flash. You don't have to wait for the spinning disk to acknowledge. So it goes to flash. And that's always been the Achilles' heel of spinning disks. So we know that well, just from understanding storage design. So specifically from a performance standpoint, what are you seeing in the field? As I say, you've mentioned NetApp several times. Can you share some other additional color on performance metrics, you know, some hard data? Sure, sure, sure. So let's take a look. There's an industry standard benchmark, the spec SFS NAS benchmark. We published this data, as did NetApp, where we achieved a 700 microsecond application response time in that benchmark. At a list price of about 500 K US, that's two terabytes of DRAM, roughly 20 terabytes of flash in the rest of HDDs. Contrast that with the NetApp Fast 6000, which the new Fast 8080X is very similar in performance, not that much faster at all. And that product, at a list price of close to 2 million, delivered a 1.54 millisecond application response time in the same test. So basically, you would have to pay four times more to get performance that is twice as slow with NetApp. If you think about that. And obviously response times are critical in these highly virtualized environments. So what are you seeing in customer environments? You've been at this now for a little bit with the latest refresh of the ZFS storage called the ZS3, where are we at? ZS3. So what are you seeing in customer sites? Where are they using it? What kind of workloads are you seeing? What are they actually seeing? So you mentioned your benchmark, what are customers actually seeing in the field? Yeah, we're in quiet period, so I can't comment on exactly what our customers are doing right now, but let's talk about what we've seen in our own environments and how that translates to customer environments. So for example, when we consolidated ZFS, sorry, NetApp Fast 6000s onto ZFS storage appliances, what we saw is 11 NetApp filers per storage administrator grow to 38 ZFS appliances per one storage administrator. We saw 500 terabytes per one storage administrator grow to five petabytes per one storage administrator. So the reflection there, the parallel is if customers migrate off of NetApp, if customers migrate their highly virtualized environments, the Oracle Database environments from NetApp to ZFS, the consolidation, the operational benefits are immense, as we talked about there. I mean, if you think about it, we went from about 50 petabytes of NetApp in roughly 2010 to over 200 petabytes of ZFS storage appliances powering the Oracle Managed Cloud Service, our IT production environments, test dev, data warehousing, et cetera. So you think about those ratios there, those are immense savings for any company. This is what you saw internally in your own data centers and I don't know if you can quantify the savings for them, but physically less equipment. Physically less equipment, quantifying. So in terms of space savings, we went down, we saved 80% by switching from NetApp to ZFS, Oracle ZFS storage appliances in terms of space, in terms of power savings, we saved 74%. So those are material benefits that could benefit any company. It's disruptive economics when you put it in that perspective. So Steve, it sounds like the strategy with the ZFS appliances, go hard after the NetApp install base, which has always had a strong affinity for Oracle, or Oracle install base that has a strong affinity for NetApp, go hard after that install base. The other thing, so Sun had storage, but somehow Sun could never get it right. So how is it that Oracle is getting it right? Larry loves spending money on R&D, I think he still runs R&D. So philosophically, culturally, it seems different than when Sun was running the base. So a couple things. One, upon the acquisition of Sun, ZFS storage appliances were shortly thereafter put into production to support Oracle's managed services cloud, the IT department. As the back end of Oracle's cloud. ZFS storage appliance is the data backbone for Oracle corporation, $180 billion company. I'm one software company in the world is powered by ZFS storage appliances worldwide in terms of our managed cloud services, in terms of our IT production department, our data warehousing, all of our application testing and development. ZFS is the benchmark. So if you think about it, growing from 50 petabytes to 200 terabytes, 200 petabytes of ZFS, right? 50 petabytes of net app, so 200 petabytes of ZFS, along the way, everything we engineered internally, the resiliency, the scalability, the high performance, building upon the foundational in-memory architecture, all of that we've transferred as benefits to our customers that tangibly translates to operational savings and performance efficiencies. The other key thing I would say that is different from Sun is obviously with Oracle database, we're co-engineering net incremental capabilities such that when you combine the ZS3 with Oracle database, it's auto-recognized, it's auto-tuned, it's auto-tiered, it's auto-compressed. A lot of vendors at this show will tell you they're great for Oracle database. They'll tell you, you know, we're integrated with Oracle database, we're optimized for Oracle database. How many, to many extents, that means they've downloaded an SDK or an API? That's very different than having hundreds of software engineers work side-by-side with hundreds of hardware engineers delivering net incremental capabilities. So within the VMware environment, it makes sense to do NFS and file-based storage, it's sort of an elevated citizen within VMware. But a lot of talk this show about VVOLs and how VVOL is going to bring the block world into elevated status alongside of NAS. So the king of block is EMC, can we expect Oracle to sort of target that segment of the market? And what can you say about that? You know, the ZFS storage appliance is a unified storage system that supports both block and file. We have several deployments, you know, lots of deployments worldwide, of customers using it for both, some using it just for block. The key point is the horsepower, it's a very technically elegant system, right? I mean, when you have an in-memory storage architecture, again, you walk around the floor here, a lot of these vendors, they're basically shoving, drives into common canisters and saying they have a storage innovation. That's distinctly different than the following. You have an in-memory storage architecture with a symmetric multi-processing operating system on top of that, with the ZFS file system on top of that and that's extracting the maximum performance out of DRAM. If you look at it that way, look at the benchmark we put out, publicly available information, SPC2 benchmark. The ZS3-4 puts out more throughput than multi-million dollar high-end arrays from the likes of Hitachi, IBM, HP. We're putting out 17 gigabytes a second of throughput, SPC2 obviously being a block benchmark. So if you look at that type of test, that type of performance, any Xan customer can equally benefit from ZFS. So we're getting the hook, but I got to ask you about tape. You have a tape system in your booth. You know what? Tape, tape is... Tell me about tape. Is tape the new cheap and deep? You hear about data lakes, data oceans? Here's the way. Is tape going to make a comeback here? Here's the way I'd look at it. We view tape as a logical extension of a highly virtualized environment. If you think about, if you have a corruption in your VM and it's on disk or flash and you keep replicating that corruption, keep replicating it, if it's stored on magnetic tape, you can go back and revert to the point in time prior to the corruption and restore it. Can't do that with any of the media. We also view tape, if you think about it, a lot of these VMware service providers, cloud service providers, you know, they're offering VMware private cloud, hybrid cloud, public cloud, they can complement their VMware environment by offering a deep cloud archive service powered with tape. Because if you think about it, you can monetize tape better than any other media, particularly for a deep archival service. So you could take all your VMware data, your VMDKs, you know, snap them to the cloud, clone them to the cloud, and you know, keep a private cloud on premise or you know, keep snapping the public cloud if you want. The key point is whether they're telcos, managed service providers, cloud service providers, anybody offering a VMware cloud service stands to benefit by offering a deep archival service for that VMware environment powered by tape. All right, Steve, thanks for coming on theCUBE Oracle, getting its act together in storage, going after the big base that is 500,000 VMware customers. All right, we'll be watching. Thanks very much. Good to see you again. All right, keep it right there, everybody. We'll be right back after this word. We're live. This is theCUBE from VMworld 2014.