 We're back, hi everybody, this is Dave Vellante and this is Oracle OpenWorld, this is theCUBE. Narayan Venkatas here, we're going to stay on this theme of Flash, he's with Violin, I'm also joined by David Floyer, David Floyer has done a lot of work in the Flash area. Actually David, I want to start with you, we're at Oracle OpenWorld, you did a piece of research last year talking about if you optimize your infrastructure stack, in particular if you spend a little bit more money on storage, generally but specifically flash storage, you're going to dramatically reduce the number of cores that you have to deploy to support applications, why is that important for Oracle customers? Well why that's important for Oracle customers is that the largest single component of any system running Oracle is the Oracle license itself, it's in the order of 80% of the cost, so if you can optimize on the number of cores that you need, if you can put in a flash as a tier or all-flash type array to do the IOs much quicker, then the net effect of that is that a whole lot of processing that you would do waiting for IOs goes away and you require far fewer cycles in the order of 40 to 50% less and that means that you can need less Oracle licenses to do a particular piece of work. Now that's not saying you're going to get rid of those licensing but what it does it reduces the cost of the whole of that to Oracle processing and that enables that same spend to go much much further and tackle new problems in new areas, like anything else when you reduce the cost then you are significantly improving, it's elastic, you can significantly improving the capability of those environments. So the whole notion of semiconductor to base storage, persistent flash based storage, it's really changing the way in which people are thinking about deploying applications, Narayan, violin, hot company, obviously you guys doing an IPO, I know we can't say much more than that but the information is now out there so that's good, people are parsing it and I know your CEO is out doing a road show so that's cool, we'll let Don Basile talk about all that stuff but you're the product guy, you're in the trenches with customers, tell us what's going on at Oracle Open World with violin. Great, so clearly Larry has spoken about moving Oracle data sets and applications in memory and so we've been preaching that for quite some time with the message that if you move the bulk of your business critical applications in a persistent memory form factor you can achieve tremendous amount of application acceleration and as a result business acceleration to David's point earlier if you look at it in totality in terms of what is the most cost-effective infrastructure to really accelerate your applications and run your databases and as we call it databases at the speed of memory, what impact does it have on your business and that is exactly what violin has been doing for the last five years which is what can we as violin offer solution sets to the customers that can help accelerate their application and help accelerate the business at economics that is far better than legacy storage particularly at scale, so that's exactly what we're doing and so what we're demonstrating today at Oracle Open World is a plethora of our solutions which are primarily persistent memory focus you know we've got solutions in the what we call memory array which is flash memory persistent storage running on the network we've also demonstrating flash or persistent memory acceleration on the server side with the PCIe cards and then we're also demonstrating a solution what we call maestro services software suite that enables you to actually pull data from applications that are running on legacy storage and help move them over to persistent memory based here so I'd encourage all of you those who are attending Oracle Open World to come by stop by or booth and you know understand the various different solutions we're offering in the market space yeah so you mentioned Oracle and 12c and in-memory databases and we heard you know Ellison last night talking about in-memory we've heard previously guys like SAP Bill McDermott has said things like imagine the world without without disk and we were joking off camera I said oh you don't have to persist to persist data anymore but of course you I'm sure imagine a world without disk at least in part so but talk about in-memory and in that trend like I say all the CEOs talking about that's a that's actually a good thing for you guys why is that absolutely so it is it is clear it's well understood amongst you know end users and practitioners that if you can actually move your application sets or the hot data if I were to call it in-memory and I say in-memory in a very broad sense you can certainly accelerate applications now we all know that the amount of data set that you typically process is much larger than your active data sets so the you know so the definition of in-memory will vary a little bit which is at the end of the day you have to persist data if you can actually move your persistent data into a memory tier as well i.e I talk about a persistent memory tier in this case it could be flash along with DRAM as as a memory tier then between the two which is DRAM as a non-persistent tier and flash or you know persistent memory tier you can actually accelerate and in significantly databases and applications layered applications so the meta-level question is if you have a broader tier of memory some of it is non-persistent and some in on the majority of that is persistent and you stay away from using disks or spinning disks as a means for persisting data then at scale you can achieve tremendous economics and business acceleration and that's what we're focused on okay now you guys made the gridiron acquisition you basically made an announcement last week sort of not only reintroducing that but but we're introducing that with some enhanced capabilities and one of those was tiering right additional tiering so I wonder if you could just take us through and unpack that announcement from last week absolutely we're happy to do that so we last week we announced a whole new sort of complement of solutions called the maestro violent maestro memory services software suite we've taken we've borrowed some technology from our acquisition of gridiron we've incorporated some of the technologies that we've been you know developing over the last several years brought them together amalgamated them and essentially set up an infrastructure where you can actually tier application data between disk storage systems that are resident on the network i.e. over a fiber channel infrastructure or an iSCSI infrastructure and then incorporate a persistent memory tier wherein you can actually truly tier the data and you can use it in multiple different ways you can use it in the form of a platform for accelerating grid-only data so think about cash you can think about a broader scheme in terms of tiering wherein the majority of your active data sets that are both read and written are actually written into persistent memory and if you really step back and think about why are we doing what we're doing given we're primarily focused on the persistent memory tier we believe this solution is an on-ramp to help you move your data sets from legacy spinning this storage architectures to a persistent memory architecture right and so I think that's where the biggest value that we see which is customers want to be able to get some benefit of acceleration be able to tier this non-disruptively and then non-disruptively move the data from a legacy storage tier into a persistent memory tier so if I understand that then you're looking at products such as the caching from netapp or the IBM or EMC fast products and saying you can put your solution in as an alternative to upgrading the systems with flash and with their software is that is that is it? Absolutely and and one of the big value propositions of the Maestro software suite is that it plugs in non-destructively to your environment into your environment right we also support a plethora of heterogeneous storage systems in the back and it sets the stage for a non- disruptive migration off of legacy storage to a persistent memory tier right now you know we all recognize that customers have got petabytes of legacy storage and they're running their business critical applications on it and and one of the questions that often comes up is how do I migrate my data how do I use my active data sets that are running business critical applications to think databases etc and how do I move that into a persistent memory tier with as little disruption to my applications as possible and what Maestro's services software suite allows you to do is to do exactly that which is get them on a path that you can tier i.e bring in a persistent memory tier to your to your legacy environment free from any vendor lock-in and then facilitate the move of the specific sets of data that are important to you into a persistent memory tier. Well you always have some lock-in don't you such as your Maestro is free for example but leaving that aside I will clarify this though I will absolutely clarify this which is while the if you if you think about what Maestro offers itself it plugs itself in transparently right once you've migrated data off off of legacy storage to a persistent memory tier you can actually take away the Maestro appliance if need be so it isn't a lock-in per se at all right I mean clearly you've moved your data sets from spinning discs to a violent memory tier right so absolutely so from that perspective you're saying that I managed to move my data sets excellent okay and so there is no vendor lock-in clearly we deliver value and and customers will will you know use us where we they think that we're delivering value in this case it's purely about how do I accelerate my our customer's business applications and can I do it add cost points that are far more appealing with our solution than it is with legacy storage particularly when you get to large scale so let's talk about Oracle for a little bit because Oracle databases tend to be the mission critical applications in organizations and people are very reluctant to take any sort of risk with that they've got their systems in place the DBAs are very very protective of their territory right can you talk about what how you would practically do that how would you convince the CIO for example that you could help them take it step by step and and preserve that environment because they're very reluctant to take risks it's a and DBAs get fired for very little mistakes absolutely so let me start with by giving you a perspective of what violence been doing over the last you know several years right which is we've been going into primary database workloads so think mission critical think business critical environments wherein we've gone in customers have used our memory array appliance memory arrays as a backend repository for their database workloads and these are highly transactional in nature or perhaps even you know focused on data warehousing but it is it is a set of applications where the the criticality criticality of that business data is very high so we've been very successful in the market already for the last several years wherein we're running those business critical database on our memory arrays right so our memory arrays are highly available they are designed for tier one workloads which means that they cannot go down there you know they've got a tremendous amount of redundancy built in so there are we've got a large number of customers who are deploying those today in production for many years so so that is a proof point of what we have we meaning violin has brought forth into this market space now we're we're starting to expand our reach which is wherein we are essentially facilitating that move of those application sets from a legacy storage environment to this right so in the context of this discussion when we're talking about maestro services etc we've built the system where it plugs in into a fiber channel infrastructure right with no change to your applications or your storage system with some minor changes to your zone in the fiber channel fabric the maestro appliance gets plugged in and you know with just a few re-scan of the pads we're able to now be presenting a highly available system along with the flash memory array to facilitate that transition right so we've actually proven already in the market space for the last several years of running business critical databases and pick your favorite flavor of oracle you know we've got oracle tan we've got oracle 11 rack all those kinds of applications being deployed already on violent memory so customers already comfortable with the idea of deploying flash well you would argue that perhaps flash was a novelty three four years ago it's become a very very crucial component of a cio strategy going forward and they're all asking the question what is my flash strategy you know how do i get my business critical applications to the fastest here such that at scale i'm a lot more economical in terms of an infrastructure and so that's what violent is out there just so that leads me to a discussion about you mentioned legacy storage before right and the legacy storage is you know putting quotes it's a pejorative basically says okay legacy storage means it means old it means maybe i i think about bolting on flash into a you know spinning disk base controller all the things that you guys like mop the floor with right okay now obviously the legacy storage vendors have seen the trends they see all the the money you know going into flash they see the hot ipos they see the acquisitions and they've been investing themselves have been buying companies so they're transforming their legacy infrastructures into hybrids so many many are doing all flash arrays many are acquiring companies that are all flash arrays so even oracle itself now has a big push with the the nfs storage of zfs storage appliance sort of hybrid so when they go to talk to executives they're going to say hey we have everything that violin has we have all the flash we have the hybrids we see the economics we're investing r&d trust us what do you say to that so at the end of the day the way you know if you go there selling the solution any solution doesn't matter right you're essentially talking to the line of business person and the meta-level question to ask is and i often ask the question if your infrastructure ran at the speed of memory and i make it economical for you to do so what impact does it have on your business if you're if your application is able to run 10x faster today today right what impact does it have on your business and so the big question to ask is do you as a customer if i'm a cio i'm asking the question do i have a competitive advantage by moving to a persistent memory infrastructure today as opposed to waiting for the next three to four years right that's the operative question and and we have a number of customers who are already doing so which is can they take advantage of a persistent memory storage system today that can help them accelerate the business rather than wait for another three years and so there is there is a opportunity at cost for waiting for technology to evolve and and and of course various different customers are in various stages of risk tolerance and we believe believe some of the more aggressive some of the more i would say forward-thinking customers i've already started to adopt memory storage right and and i think i think the meta-level question is do you want to wait several years or do you want to take advantage of it today so you said three years you feel like you have a three-year advantage on the legacy storage we've been we've been doing this our belief is that the the active data sets will transform or will change and the infrastructure will transform to memory data memory but it's completely great so we're not saying that while you're going to go put your tier three or your inactive data sets in memory because it's not economical right so what we're out to do is to actually bring the acceleration or help customers move their active data sets today and and there's clearly you know there's an opportunity cost of waiting right why take that you know why take the risk of waiting right excellent darian well thanks very much for coming on the cube and uh good luck with all the action in the marketplace good luck with the ipo and uh really appreciate your time great thanks david thanks day for having us all right keep it right there buddy we'll be right back with our next guest this is the cube we're live from oracle open world