 We're back, this is Dave Vellante. We're live at Oracle Open World. If Moscone is draped in red, it's Oracle Open World 2013 and this is theCUBE. We're going to continue on with this discussion that we've been having around Flash. Gary Ornstein is here as senior vice president at Fusion I.O. and a longtime CUBE alum, Gary. Good to see you again. Thanks for having us. And of course, David Floyer is with me, resident Flash expert, infrastructure, database guy at Wikibon. So Gary, we were last here at VMworld talking off camera. It's a totally different vibe here, right? I mean, it's like big and everything's draped and you feel like you're inside the tent. And at the same time, Oracle's now claiming 60,000 people. I think they're doing that, David, because Salesforce is claiming 40,000, so they've got to be bigger than Salesforce. But so what's been new since VMworld, Gary? What's happened in the year world? So we have a number of things that we're showcasing at Oracle World around the use of Flash memory and Oracle. And it basically boils down to three things. We want to help customers do more transactions. We want to help customers make faster decisions and we want to help them save money. And all of those things are enabled by using Flash memory compared to conventional architectures where people were relying on disk drives for performance. So we're sharing a number of those things here at the show today. Also integrations that we've done with Oracle Enterprise Manager and Oracle Virtualization Software. But to make things simple, it comes down to more transactions, faster decisions and at a lower cost. And that's a message that really can help any customer. So I mean, I just make an observation and wonder if you could comment. It's like, whatever you hear in the marketplace with the trends to come up from the startups and the innovators one to two years ago, you then hear at Oracle Open World. It really, it's true. We heard today about key value stores. We heard about Hadoop connectors. We heard about bringing analytics and transactional data together. And I say that sort of tongue in cheek, but the reality is, is Oracle goes after the fat middle. And that's actually a good trend for you guys, isn't it? Yeah, I think it's no surprise that we knew there would be a big focus on in memory at Oracle Open World. We've been talking about memory for quite some time. I think one of the nice things about hearing Larry talk about it yesterday during his keynote was the need not only to increase the number of transactions, but the need to increase the query time and really to reduce the query time, to increase the performance or reduce the query time. And we've seen this time and time again in customers who are looking to do essentially, many type of things simultaneously. Either they're looking to increase transactions and increase queries. Another common workload that we see is the need to ingest data, the need to process that data, the need to present that data in a way that's actionable for the decision makers, and then a way to archive that data to a storage medium that makes sense for long-term archival. Doing all those things simultaneously, ingest, processing, presentation, and archiving. You can't do that on conventional infrastructure and you have to move to a memory-based infrastructure. There was a lot of talk about DRAM yesterday. DRAM is a great solution for a lot of people. We happen to think that flash memory provides a better cost performance balance for folks who are looking to process these kind of transactions, and so we've focused on providing our customers a way to do these type of things with a flash memory infrastructure. Well, it helps that it's persistent. We've heard guys like Bill McDermott say, imagine a world without disk drives. Well, I'd love to imagine a world without disk drives, but I can't imagine a world without persistent storage. Right, so you're always going to need that persistence and flash memory is fitting a really great middle between DRAM and disk drives to allow customers to transact more, decide faster, but ultimately save more money. One of the things that I was pleased to hear Mark Hurd talk about is last night Larry Ellison kept talking performance, performance, performance. This morning you heard Mark Hurd say, anytime a customer wants to take a portion of that performance and allocate it to cost savings, they can do that too. We shouldn't forget that the numbers that we increase on the top line can be flipped around so we save on the bottom line as well. And when we're talking so much about flash memory and performance, we sometimes forget to remind ourselves and also our customers that there are huge opportunities in cost savings and shrinking the infrastructure with no sacrifice to performance whatsoever. So I'd like to ask Gary one question about this in memory trend that we've been seeing. Clearly the requirement there is to get analytics in particular into real time, as much as real time, close to real time as possible. And obviously clearly you can do bigger transactions, larger transactions if you have the data memory for the transaction systems. So it seems to me that in that environment, the requirement for flash is actually very, very important because if you have to put that data out onto a disk drive of any sort, you've got a total bottleneck for recovery or anything else like that. Is that what you found in practice? Is that what you observed? Sure, some of this is always better conveyed in the words of our customers. We have a customer who said, if I'm running my application and I'm serving my end users, and I have to leave the motherboard in order to complete a transaction or to complete a query, I'm kind of in trouble at that point because the speed delta is literally like falling off a cliff. And so by using flash memory and having it be right there in the server that's processing the application, they're able to essentially achieve that real time workflow, which is so critical. We live in a world where time is of the essence and nobody wants to wait. And you combine that with the proliferation of mobile devices and endpoints and the need to serve all of that stuff from a single data center, a consolidated application from single data center, and there is no room for anything that's not in real time in this day and age. One of the other things that we heard this morning, I don't know if you're able to catch the interview with the CEO of the NYSE. He said that we used to measure the transaction time in seconds just a few short years ago and now we're measuring it in microseconds. David, you've done, I remember when Fusion I.O. came out with the benchmark of a billion I.Os. You actually did a hypothetical model of how many disk drives it would take to achieve that. It was just a phenomenal, an acre of disk drives that were required to get to the same level. So to Gary's point, right, you can always get the performance. It just really wasn't practical to get to the performance. It was $30 million, I think it was, to achieve the same level of performance that was achieved with a half a rack of equipment at the demonstration. So Gary, as always, it's all about the application. So talk a little bit about how your customers are transforming their application portfolio. Sure, we've seen a tremendous range of applications that customers are using for Oracle. Of course, Oracle is the foundation layer upon which they'll do a particular application. So just yesterday we announced some of the use cases of our Ion Data Accelerator. As an example, Ion, the Ion Data Accelerator is an all-flash appliance, but it's built on servers that our customers know and trust. So pick your favorite server from HP or Cisco or IBM or Dell or Supermicro and you can run the Ion software to share flash memory. So some of the examples, one of the world's largest shipbuilders has their entire product lifecycle management database for all the shipbuilding parts. They're able to create 3D models of what needs to be done in 15 seconds, what would have taken the minutes or hours previously. We have one of the largest grocery chains in Asia that's recording point of sale information on Ion. We have companies that are deploying analytics and shrinking their query times from many hours down to minutes so they can make decisions faster. One of my favorite stories is customers who are able to essentially create a new business day. So what they used to do before Fusion Ion came along is they would wait until the end of the business day, let's just call it five o'clock, and then they would run their queries because it would take many, many hours, sometimes running into the night and then they'd come in in the morning and they'd be able to assess where things were, supply chain, inventory, demand, and make a new set of decisions. Now with Fusion Ion, they've shrunk that time to just minutes. They now run those reports at 3 p.m. in the afternoon. By 3.30 they're finished. They've got 90 minutes left in the working day to act on those decisions. It is literally creating a new business day when it didn't exist before. So that's the kind of transformative change that we see with our customers being able to really do things that weren't possible before enabled by either using flash memory directly in a server or using flash memory in a shared configuration such as with our Ion data accelerator. So this is why I always like talking about, talking to Fusion Ion because everybody always talks about the economics of flash and oh, we shouldn't be talking about cost per gigabyte. We should be talking about cost per Ion. It's really not even about that. It's about the business value. And David Floy, you've made this point a number of times. Absolutely. And in fact, one of the business cases that we looked at was at Revere, which is a small outfit in Chicago, was actually running the whole of their database on a Fusion Ion system. And what they found was that by doing the very thing you're talking about, being able to run all of the software at the same time, they could do the queries, they could do the purchasing while other things were going on, while the call center was going on, that enabled them to increase their revenue by 20% without any headcount over less than a year. And the return on that is as 20% straight to the bottom line. The return on that is another way of thinking about the whole benefit of speeding things up. Yeah, I think time is money. And so when we're able in that case to allow somebody to run their transactions that they need to do on a daily basis to keep their business afloat, plus run the analytics so they can tune their business and adjust appropriately and do that at the same time. It gets back to the same comment I made in other cases where people need to ingest, process, present, and archive that kind of flow at the same time. In this case, people need to be able to do the transactions plus do the analytics at the same time. So all of these things come back to essentially creating time opportunities for our customers which translate into real dollars. So Gary, we're talking a lot about Fusion I.O. and I always love having you in the queue because we can talk about a lot of different subjects. You got a great, you know, you're part analyst. You know, you were obviously a fairly prominent journalist and obviously now a technologist. So I wanted to get your take on open source. You guys are big in open source. You've made some investments there. You've supported the community in a big way. What's your take on what's going on in open source specifically within Oracle? And, you know, a lot of people are saying, oh, the open source community will catch up. You know, Oracle's really not committed to open source. It risks their business model. At the same time, there's a company with 39% operating margins, 45% last quarter, throwing off $14 billion, you know, and 12 months of free cash flow. So something's working, you know, but the naysayers will say, and I've been one of those, that geez, the open source community will eventually, you know, catch up and provide equivalent value to the mainframe, the red mainframe. What's your take on all that? Yeah, I think there are certainly opportunities where people may have been using one commercial application where there is a suitable opportunity to use an open source application. So, you know, we see a lot with the new popular data stores, such as MongoDB and Cassandra, which seem to be at the intersection of adoption, but also budget being assigned to those applications. Of course, the big story in open source around Oracle is MySQL. You know, we don't see as much about MySQL inside of Oracle as we saw without Oracle. We, at FusionIO, see a huge opportunity with MySQL, and I've done a lot of work there to do optimizations that go above and beyond just providing acceleration from flash. Specifically, I'm referring to the ability to do atomic rights, which provide more performance and less infrastructure and extend the duration of the flash memory. What you've done with Percona. We've done that with both Percona and MariaDB, so those for customers who are using MySQL from Percona or from MariaDB, there are phenomenal opportunities with them in the open source arena. The open source arena, I think, is going to continue to provide the innovation that is needed to showcase what's possible with these new technologies, so we're focused very much on evangelizing the capabilities of flash-aware applications. We're still in a world where many applications were the thinking behind the application all funneled down to, how am I going to optimize IO? All these requests are going to funnel down into a single platter with a single moving head on it. That's quite a different world from where we live in today where you could throw thousands of simultaneous requests to a flash memory product from Fusion IO or elsewhere and have those responses all come back at the same time. So, this is not something that's going to happen overnight. We're seeing the increase in awareness of flash-aware applications, but it will take time as with all new technologies. But I do think that open source is in a very unique position to move quickly and showcase what can be possible, not only with open source applications, but also with commercial applications too. Gary, thanks very much. I always appreciate your perspectives. We are running out of time here, but love having you on theCUBE. Welcome back any time. One more quick thing. I mentioned that we have a great demonstration at the Fusion IO booth in conjunction with Cisco and Oracle and Emulex. We're showcasing a Cisco UCS-based Oracle configuration doing 2.5 million IOPS and that's from the Oracle Orion Storage Benchmark. And so again, this is an opportunity take the performance up at the top, take the cost savings down below, but if you want to see the latest and greatest in terms of Oracle performance or cost savings, I'll swing by the Fusion IO booth. Awesome, I appreciate you coming by. Thanks very much. And keep right there. Teradate is coming up next. We're going to unpack what's going on there with one of Oracle's bigger competitors. Keep right there. This is theCUBE right back after this.