 Okay, welcome back everyone. I'm John Furrier with SiliconANGLES. This is theCUBE, our flagship program. We go out to the events, extract the signal from the noise, and join my co-host Jeff Frick with theCUBE. And our next guest is Nisha Talagala with Fusion I, a lead architect here at the show at Percona Live. Welcome to theCUBE. Thank you. So we always talk about the software defined data center and that has been the rage since, I think VMworld two years ago has started to see software defined networking, virtualization. That kind of gets everyone's attention to the changes in the data center. And that's great and all. We'd love to see this software modern era kind of evolve. But one of the things that we've been following with Fusion IO is just how well flash is played in the optimization of data center performance, speeds and feeds, IOPS if you will. But now that's all evolving, right? And you guys announced yesterday the nonvolatile memory compression, Oracle supporting a big endorsement that came in late in the wire last night. We saw that. You gave a talk here on the data center, all flash data centers. So that's a big battle ground for many levels, not just companies trying to get market share, but for customers looking at how to evolve and be positioned for the future in the data centers. Share with us some of the things that you see and highlight some of the things you talked about this morning. Sure, so I think the transformation of the data center through flash in the last few years has already been very, very substantial but it's still at the very beginning of this particular journey. And so what I covered in my talk today was the overall path of evolution of flash and the data center seen from the point of view of the MySQL database as a specific application. And there's been a number of different, very substantial things that have happened in the last few years, mostly about adapting to the speed and the sheer speed of IOPS and latency that flash brings. And a lot of great changes and optimizations have come up in the MySQL releases of the last few years to really exploit that. What we announced yesterday, which we were very excited to be able to announce with basically all of our partners, MariaDB, Oracle and Prokona was what we consider to be the next wave of optimization and enhancements for databases with flash. And that is what we call flash awareness where you're recognizing the fact that you're not dealing with a really fast disk necessarily, you're dealing with a brand new medium, medium that is now pretty much a staple in the data center and which is why we use the term all flash data center. And then now how can we adapt and improve applications to be able to make use of that. So that's what we mostly announced yesterday. We announced a feature called NVM compression that comes on the heels of a previous feature we had announced called atomic rights and we're very excited about it. You know we always talk about the history of the computer revolution and disk used to make up for all the inadequacies of RAM and page swapping out and all kinds of memory management, device management stuff. Now it's the other way around. Disk is now the issue and when you still look at some of the non volatile memory performance like looking more like DRAM. How is that changing the game for developers especially as they look for the scale not just scale in terms of performance but now you have multi platform environments you deal with a lot of different kinds of stacks. Yes. Talk about that dynamic. So I think there's a lot of different things. So one of them is that in the disk era world developers spent a great deal of time trying to figure out how to overcome some of the issues or limitations of disk drives. And so things like how you lay out your data and being very aware of the fact that there's a disk arm and things like that. There was a tremendous amount of technology and software that was developed just to optimize for that. Flash renders most of that largely unnecessary because flash is so performant that a lot of the difficulties that you had to go through you don't have to go through those anymore. It opens up a new set of opportunities. You can certainly consider them challenges or opportunities but opportunities in the form of there's so much more parallelism now. How do you make use of it? Can you write software in a way that's simpler more scalable that takes the best use of the underlying flash? And this is again, while this is transformative in itself it's also the beginning of a larger trend because flash is much faster than disk. It's much closer to memory. The ability to see it in a more memory context is far greater. The technologies that are coming down the pipe are even lower latency. So therefore there is this, what we call convergence of memory and storage where the line between DRAM and disk was always such a hard line but the line between DRAM and flash is very blurry. So I wonder if you can give a couple of examples of developers using the new capabilities in the all flash data center to improve existing applications. But then I think what's more exciting and I think what Gary Orson talks a lot about is this infrastructure enabling really a new way to think about the problem. And if you can give some examples contrasting to that type of net new opportunity that's kind of flash from the bottom up in terms of the application. Absolutely, so I think, so one of the things I covered in my talk this morning was sort of the three phases of flash evolution. So enabling these technologies to benefit existing apps is what we've called flash as disk, which is it is a really, really fast disk. And so you can get a lot of benefit just by seeing it that way without having to become flash aware or anything like that. And so you see applications running many degrees faster, being able to deliver more transactions per dollar per watt, which is the most important thing. So it's not just that it's faster, it's actually more efficient for a unit of work done. So all of that has been going on for the last five years with tremendous success. And that has propelled the more broad, widespread adoption of flash. What we're announcing is sort of the second wave, which is flash awareness. And that can be applications adapting, it could be file systems, for example, adapting, and thereby any application, even a legacy application, that runs on top of an optimized file system will benefit. So one of the reasons why we've developed our own file systems, it's called NVMFS, is to enable that kind of benefit for unmodified apps. And then the transition into much more of a memory oriented world is a third step. When does that happen? I think it's imminent, probably somewhere in the next few years, we will start to see that. I mean one of the things that we introduced last year as part of some of our open sourcing was enhancements to the Linux kernel to improve virtual memory. So the first phase of that is already happening. Some of those changes have already made it into the upstream kernels. And what that does is it changes the virtual memory system so that applications can still think they have a lot of memory. But what they really have is a tier of DRAM and Flash. And so they can benefit very transparently. So it's already happening, and it will ramp. Nisha, talk about some of the applications that are coming around the corner from this. Massly, this creates a different dynamic from a programming standpoint. We were talking on the opening segment about MySQL's challenges around out-of-the-box replication and high availability areas that you guys are very comfortable with and have done some work in. How does MySQL go to the next level? And as things roll out in the data center with the memory architecture changing and evolving, what are some of those new applications? So how does MySQL scale up out-of-the-box in the future with replication and high availability? And what are some of the new applications that we see come around the corner? Sure, so I think it's one of the panel this morning actually spent a lot of time talking about the future of MySQL at scale. And there's a lot of people who are already deploying MySQL at very large scales. So I think there's a lot of learning, a lot of practical deployments in companies like Facebook, Twitter, the folks who created the web scale, Alliance and so forth. So all of that is happening fairly naturally already. As far as trends that are coming down the wire, one of the biggest trends that we see is this memory and storage convergence trend with some of the in-memory databases. And so what people are finding is that the classic world of my analytics is over here, my transactions are over here and they don't meet and things like that was in some sense created out of necessity because you couldn't put the two together. It just didn't work. But now when you have these faster technologies, suddenly I can have my transactions, I can have my analytics be up to the moment and things like that. And people are responding very much to the value of that. So that's probably one of the biggest trends. We were talking about last night's event we had up in San Francisco, AT&T Park with all the sports teams, big data and sports kind of intersecting. So it was more of a fun event really with the executives. We weren't getting in the weeds like here but they're talking about vision, leadership and big data has focused the attention on using data first. Really designing around data as a competitive advantage and why we're so excited about the web scale SQL announcement is it really points to companies like LinkedIn, Twitter, Facebook, Google who use data as a competitive advantage as part of the core part of the application, part of the development process. So with that, what are some of the things that you're seeing architecturally that companies are doing or trying to do to be data first? And how does that compare and contrast just even a few years ago? So I think it's being data first absolutely completely agreed. It is a very, very big trend. And some of it that is, how do I make sure that ways that I process my data have the most up to date data? So if data is coming in from one source and being analyzed in another source how do I put them together as quickly as possible? How do I best use some of the upcoming innovations and processors to generate as much processing of the data as possible and things like that. And so we're seeing that in the database community we're seeing it in the SQL community around some of the analytics. We see it with products like SAP's HANA and things like that. And so it is everywhere. And most of it is about low latency, efficiency and how to make the best use of your resources. What's the next big challenge that you're looking to overcome? I should say this community is looking to overcome. I think an interesting question. I mean, I'm not sure if there's necessarily one challenge but I think the quest for efficiency and scale is pretty much never-ending because particularly when you're deploying things at massive scales even small inefficiencies in a single node translate into millions of dollars of operating expenses. So I think there's an ongoing shift to how do I squeeze expenses out because most people, the money that they make is between the price of the service they offer and what it costs to run that service. So the quest to increase efficiency is just ongoing. It's always been a challenge but I think it's still a challenge. It's still the really kind of incremental takedowns little by little by little by little. Jeff, I got to ask you a question, Jeff, because you're more looking at the business landscape. What's your take on what people are thinking about in the data center just even last night some of the other customers we've talked to? What's your take on the business side? What's the mindset? I think as we move closer and closer to kind of a rent versus own world it just seems to make more and more sense to outsource the data center operation to people that specialize in data center operation just like the great service that we all use forever, ADP. Nobody ever talked about ADP. It's like the great first outsourced internal operation that nobody does anymore. So again, why would anyone want to run a data center unless it's their core competency? It's complicated with rising energy costs, changing infrastructure and architecture. It's hard to keep up. So why wouldn't you put as much of that out to a specialty provider out there as opposed to doing it yourself? Really concentrate on your core competency. I think we just continue to see an evolution of people focusing their internal resources on the things that give them competitive advantage versus infrastructure that they can put out to somebody else. I just love ADP because nobody talks about it. They've been doing this on the payroll side forever and everyone knows what exactly that means. Neisha, I want to ask you about the show for the folks out there watching or watching on demand. Why is Percona Live an important show? I mean, it doesn't get the fanfare that Microsoft build gets right now since the Windows phone, everyone's talking about it compared to the iPhone. But why is this show important right now? So I mean, I think this show is one of the biggest opportunities for the MySQL community to kind of get together and talk about the community itself, the features, the software and things like that. And MySQL is a very, very important application. All of the distros, MariaDB, Percona, Oracle MySQL, very important application. We have a lot of customers who run it in the messaging space, in some of the gaming space, even in the enterprise spaces. And so I think it's a great show. There's lots and lots of vibrancy in it. What's been the biggest surprise? Either upside or downside? I don't know that I've seen a downside. I've been very happy with the show so far and how much, in some ways, the show talks not just about the technology itself, but also about the process. How is the community doing? Is it encouraging everybody and things like that? I thought that was great. And how is the community doing? Because it's one of the older open source communities around. We go to a lot of open source shows, open networking, some that we were at the other day that there's a lot of newer open source projects, open stack summit we'll go to in Atlanta here shortly. But this one's been around for a long time. So there's probably unique challenges that come just with being a little longer in the tooth than some of the newer open source projects. From what I've seen, not just in attending the event itself, but just the experience that we've had working with the partners, it's been very positive. I think the pace of change is really great. The pace of innovation is really great. People adopt things very quickly. So I think it's very healthy. Oh, good. So I want to ask you a final question. Give us an inside look at what's happening at Fusion IO from a product standpoint, from an execution standpoint. So you guys have done a great job as you guys go into the data center. Do you see the Software Defined Data Center really emerging, and what are some of the things that you guys are doing right now as a company and product? Sure. So I think Fusion IO's product portfolio has expanded very substantially in the last few years. So the company was initially very successful with the PCIE flash devices. That is really where we established the market. That is still a very, very significant part of our business. But some of the new things that we've introduced in the last couple of years, particularly our caching products, things that we've done in the VDI space with the introduction of IO VDI at the last VMware show, as well as our appliance products. We have two different storage appliances now beyond the IO VDI, the IO control, which is a low-end hybrid kind of storage for more of the SMB market, as well as the Ion Data Accelerator, which is for the higher end all-flash market. So we continue to sort of expand our product portfolio and focus on application acceleration at all levels. So we're here live at Prokona Live. We'll be covering the rest of the day here on siliconangle.tv. I'm John Furrier with Jeff Frick. We'll be right back with our next guest after this short break.