 Okay, we're back here inside the CUBE Silicon Angles flagship program. We go out to the events and extract the signal from the noise. I'm John Furrier, the founder of SiliconAngle.com and we are here, live in New York City, where it is big data week and we are expanding all of our coverage to look at the business value of big data, looking at the analytics side of the insights and then under the hood, looking at all the technologies from ingesting all the data, managing all that, computing calculations on that data to ultimately looking at the impact of the business and vertical markets and insights, et cetera. I'm joined here at the Fran Kahn's SBP, Chief Technology Officer at Sybase SAP. So welcome to the CUBE. Thank you so much, pleasure to be here. So we love SAP, the CUBE started in 2010, it's starting our third season of going to events and we've been at three straight sapphires where we've covered SAP in multi-day format like here in the CUBE and watching SAP evolve was awesome and then the second season of the CUBE, they made the big side-base acquisition which created a lot of conversation but at the time was an early signal of essentially big data. And last year at Sapphire, I had a chance to talk to Bill McDermott and Jim Schnabe about big data and SAP kind of looked at me like, well, yeah, I'm like, you're a big data company SAP, you have side-base, you have in memory, you have mobility, you're showing analytics. I mean, it's the poster child of everything that's being talked about here in terms of business value analytics but SAP wasn't taking any credit for it. So I asked Schnabe, I said, why don't you just market around big data and his reaction was in a very German way, well, it's kind of hype-ish. So SAP, a very professional company, so I understand that but truly it's a big data company. Can you explain to the folks out there what I just said in terms of how SAP really is a big data company when you look at what the startups are patching together here in the open-source community, you guys kind of have that. So explain to the folks out there what SAP has and include the side-base. Sure, so firstly I'm appreciative of you acknowledging the big data credentials of SAP. I mean, it's worthwhile pointing out, of course, SAP is one of those stealth companies in terms of the impact that it has on the world's economy. I mean, if you take a look at SAP ERP, for instance, I mean, it's a pioneer, of course, behind the whole packaged application movement. More than 60% of the world's GDP will at some point in time touch an SAP piece of application infrastructure or applications per se. So that in itself gives it a massive amount of spread in terms of the pervasiveness of SAP. Now SAP over the years has driven itself by certain standards, okay, and those standards are enterprise grade. So whereas a lot of the startups here, and it's fantastic seeing all these companies congregate in this location, but a lot of them are quite somewhat new to the market. They're trying to find the feet, they're trying to figure out exactly where they stand. SAP, of course, has, you know, is tread the path of a school of hard knocks, so to speak, okay, right? So it knows exactly what it takes to have enterprise grade and mission critical support. So SAP in a nutshell from a big data perspective, A, has huge levels of systems of data or systems of record rather, storing business processes that have been around for many years across many lines of business and across in the classic ERP. And now with the SAP acquisition, A, they get an acceleration into mobility, but B, also some fantastic foundations in database and database technology. And that wasn't really talked about much during the acquisition, although, you know, on the inside the ropes, you know, people understood that database impact, but talk about what's changed with SAP, with SAP, side-based with SAP, in the database area specifically, because since the acquisition, you've seen the explosion of open source scale out, H base, you got MongoDB, a bunch of variety of other databases cropping up. And we've been saying on theCUBE here, it's not a mutually exclusive situation. Other companies like IBM and HP recently, we were on this morning, we were at IBM earlier in the week at IOD, and they're embracing the Hadoop and the early emerging ecosystems, and they're putting their solutions on top because they have that enterprise grade kind of level. So talk about the role of SAP with side-based, just over the past 18 months, what's changed, if anything, and how have you guys integrated in considering this thermal growth of emerging open source big data stuff? Sure, it's a great question. So if you look at side-based two years now, since approaching two years since the acquisition, left somewhat independent, but then at the same time supporting the parent company's desires for business applications, right? So firstly, the transactional database engine, the flagship database engine of side-based is ASC, Adaptive Server Enterprise. And over the course of the last two years, that's migrated along the lines of not just supporting custom applications, but also now supporting business suite applications as well. But broadly speaking, where SAP is now leveraging side-based is really the technology know-how, 25, 30 plus years of technology know-how in database and database technology. So where we have now the flagship product set of the database company SAP, if I refer to it as a database company now, is really along the lines. Thanks to you guys. Exactly. Along the line, actually, even independently. Yeah, they have some database, yeah. The real growth really in terms of innovation and market opportunity is through this new innovation of in-memory. Now, SAP has made substantial investments in in-memory compute. We have a product, of course, called HANA, which has not only taken the market by storm, but also a lot of the traditional vendors out there who are now looking at SAP thinking, well, where do they suddenly come from? And now they're having to almost scurry around to try and rebrand themselves as becoming in-memory database companies as well. So I think side-based, over the contributions of the last two years, supporting business applications by having a first-class database offering underneath those business apps, and secondly, contributing towards the innovations, kind of bi-directional innovation between in-memory computing. Side-based is know-how in that space, helping SAP evolve its own platform offering in in-memory computing as well. So you guys have been ahead of the curve, and we recognize that. So two years ago, with mobility was the big theme at Sapphire and with SAP, and Shnabe talked about the growth strategy, and then the next year, you've got success factor. It was kind of bolted in at the last minute, but so then it's clouds. So you've got mobile and cloud. Now you have social and you have connected internet of things. There's a whole other new era. I'm sure next year, Jonathan Becker will have some sort of new marketing angle there, but okay, talk about what that means, because now you have to integrate it in. You have business by design, these other environments where cloud is critical, but it's kind of an older definition. So the question is, not that it's wrong. It was implemented kind of before this kind of disruption was happening. What is the current plan for the cloud and the big data as emerging techniques that look similar are emerging? So I mean, first and foremost, I mean, we always are too quick to literally write the obituary of everything that's preceded big data or whatever the trend is of the day. Yeah, that's for sure. Frankly speaking, I mean, SAP strategy is driven on three levels, right? So firstly, there's an on-premise business, which will be around for many, many years to come. There's an on-demand business, and the on-demand business really is heading in that sass more importantly on the cloud-based side of things, and of course there's the on-device business, right? Which is a whole mobility drive. So across all of those modalities of on-device, rather going backwards, on-device, on-demand and on-premise, we are putting equal levels of evaluation and value into the business to evolve. So you said success factors becomes a cornerstone behind applications. HCM, human capital management type of things within the cloud. As far as SAP's underlining transformation is taking place in the database here, HANA, which is our memory compute platform, really beyond what we can actually even articulate today because the market really hasn't quite caught on yet. Internally, we have the buzz, we know what it's able to achieve. Give us some taste of some of the benchmarks. We've seen some numbers at SAP, I want you to share some anecdotal performance increases because I've seen things that I've heard numbers saying five minutes, five seconds, five days, five minutes. So talk about some of the order of magnitude, performance gains that HANA has been implementing. Before I do that, the anatomy of what HANA as a platform provides you with. First and foremost, it gives you a unification of OLAP type environments, so the analytical type environments, but also OLTP. Now, some of the OLTP testing, in fact, last week, TechEd in Las Vegas, we announced some numbers based upon some performance benchmarking. We're able to achieve 1.5 million transactions per second using HANA as a persistent store with solid state devices, of course, but 1.5 million transactions per second in HANA. Now, if you combine that with some of the analytical capabilities, you're able to minimize the amount of latency that you incur between a transaction occurring and then being able to do very deep analytics, very deep analytics where the data actually resides. So it's phenomenal, it's not just driven by the pure transaction performance, it's also driven by the levels of insight you can drive. I kind of test that you guys have done a great job on the performance side, and I love the strategy of SAP. You got to say, I think you guys are very relevant, and that's why I say I think you were big data before, actually built out big data value proposition and executing it at scale and grade, kind of before the whole market, but again, that's a whole different discussion. The consciousness we want to shift to now is in this ecosystem of Hadoop, you guys have some announcements, but relative to this emerging marketplace, how do you guys look at integrating in some of the Hadoop stuff? I mean, you have deals, have you announced some deals with Cloudera? Can you just go through some of your presence here at the ecosystem? Sure, so given, as you say, SAP is not exactly a new market entering into big managing big data, some of the assets that we have from the Sybase side, we have Sybase IQ, one of the first column based databases in the market, there are three philosophies that we held for IQ in terms of integration with Hadoop. Number one, you can have the classic connector, so Cloudera has connectors of course, we can have the pool model, being able to inter-operate with Cloudera from an ecosystem perspective. Number two, we have the ability to be able to actually federate our access to Cloudera HDFS type file systems and be able to map tablespaces across to them. And the third one is, is actually being able to use the database itself to be able to push processing down and actually do native map and reduce using SQL-like extensions natively within the database. So native in database Hadoop type processing. So our extension really is, is to work with the market incumbents, the big market leaders, Cloudera, Hortonworks, et cetera, but also build the bundling so the customer's not left with all the heavyweight lifting. Drive that innovation in terms of the interoperability. How is it that you can get, say the business objects data services? We have this ability to be able to farm, federate down out type of queries using business objects as well. So we have a very comprehensive bundling, packaging, but also a technology grant as well. In the last few minutes we have here, I want to talk about some of the, some of the in the weeds questions around connectors and other things and benchmarks in particular. But let's talk about connectors. The strategy for Hadoop for most people, hey we'll build a connector and suck data around and move data around. And that's, I guess okay if you're moving it and you have, you know what you're moving it for, but it's not the answer. MapReduce in the data sets themselves as seems to be the platform of choice. So do you agree with that statement? And how does that relate to SAP? Yeah, I think the question really comes down to the specifics of the implementation. So yeah, at the moment, because these things are being conceived as almost bolt-ons to existing infrastructure, you're having to deal with these connectors. But ultimately the goal should be to push processing as close to the data as possible. Now if you're going to have to federate out every single time to get to a HDFS or Cloud Air or whatever Hortonworks, you're going to minimize or rather you're going to incur more overhead. So overall, you could argue and say, well if you could well petition the workloads and have map and reduce type functions in the store itself, that would be the most optimum. And in fact, from an SAP point of view, we're driving towards that. We're driving towards having the ability in, in for example, a HANA-based platform to segregate the workload and being able to do the unification of all types of workloads within one single store. The word unification is being kicked around again. It's the tech industry's favorite word. You know, almost as good as big data, unified communications, unified storage, unified big data. We can keep watch that buzzword and that hype, but it's really relevant. But okay, on that point, I agree with you by the way on the whole, moving processing down to the data. And a lot of other people do as well, so accurate there I think. But benchmarks, so here at the show we had an entrepreneur come on from Aerospike, Brian Bulkowski, really smart tech alpha geek, and he's got this unique in-flash database. And we're talking around like, who's got benchmarks? So no one at this show yet has released any kind of framework for how to think about performance. And us, you guys nailed this with HANA. Brian's company Aerospike has nailed it. MapR's did his TerraSort. Apparently no one else showed up so they're the world champions. I saw that 58 seconds or whatever. 58 seconds. TerraSort's interesting, but you know, it's good. But no one else is doing any benchmarks. The question is to you from knowing what you know at SideBase and what you've worked on and what you manage, you know, you're managing a lot of big iron, big software, and distributed software, all that great stuff. What benchmarks should this industry be looking at for big data? Benchmarks in themselves, I think, even traditional benchmarks, the TPC type of benchmarks, the Transfer Action Processing Council, they are effectively all workloads that are driven by a certain profile of user and a certain profile of transaction. Now, you're absolutely right. There is no in-memory specific benchmark. There's no in-memory specific workload right now. You end up having to balance out, well, what are you exactly trying to do in memory? Massive sorts like the TerraSort, or is it going to be huge levels of applications who are going to be doing OLAP type functions in memory? So I think from an SAP point of view, we actually have laid down the foundations for an in-memory benchmark. We've created an in-memory benchmark. We've tested HANA against it. Well, you have use cases. Precisely. So you're saying here there's not many use cases yet emerged here other than in-memory. I would say that the use cases ultimately will drive the workload of those benchmarks, but frankly, I think TPC needs to step up, perhaps, and actually author some of these pure in-memory kind of benchmarks. At the moment, you have this traditional go-to disk, bring the data back, and then the whole concurrency piece that you have to worry about from a benchmarking point of view. It's somewhat contrived. I think what we truly need to get to is use case-specific benchmarks that allow you to be able to truly exercise what the engine is able to do. We were just talking about another AlphaGeek, and it's cool to be a database guy now, and one of my tracks in computer science degree was database, I never told anyone I was in database business, because it's like no one was falling out of their chair, but now it's cool to be in databases. So we're talking about concurrency, that it's seeing it's a transition in concurrency. Do you agree with that, that it's aging fast and not aging well, and that new kinds of concurrency is needed? So if you think of the typical model, it is transaction workloads are updating data, so therefore the update model incurs massive concurrency overheads. If you go to an append-only model, this is exactly what we're driving with HANA. So the notion is really that you think of the hot temperatures of data, so hot, warm, cool, and then cold, and over a period of time, daily when it gets cold, you essentially archive it off. But the problem with that is, of course, is that you need to then bring it back to hot data again, some compliance check, for example, some regulatory report needs to be run. We have Glacier, which is Amazon's product, actually call it Glacier, it's actually completely frozen. So the intent really is that, if you really look at the profile of data, you need to be able to get to the point where you can bring back data into the data. Isn't that why some of the things are changing so fast, because what you just explained is kind of a new phenomenon. It used to be simple, active data, passive data archive. Now you have different classifications and you need low latency. And is that one of the drivers that's changing a lot of this? It's going to become increasingly important to be able to balance right and read optimization in one single platform. Now I don't use that word unification half-heartedly. We have actually got a very strong, credible platform, which is not encumbered by this technology debt that's been carried by a lot of the other vendors out there, they've carried this through because they've had to incrementally add new features to their database technologies. We've started afresh, we've worked with Intel, we've innovated in in-memory compute, we have a foundational platform, which is very CPU cash aware, and our in-memory platform is built around that. You talk about concurrency, we've actually introduced new concurrency rules as you talk about that, not just classic MVCC multi-version concurrency, but also being able to go beyond that so you can look at the different dimensions of data. So does that take into account Flash? Absolutely, I mean Flash in one, it's just another medium of storage. Yeah, you look at it as a storage. No changes to that because of Flash. It just accelerates it. Some vendors would argue and say that's the panacea to all performance problems, but of course they're still encumbered by this technology debt. It's nice, it's changed the game, no doubt, but it's not going to change anything with that. I mean, ultimately you're still playing catch up. I mean, because if you don't address the root cause, which is inefficient algorithms, in those database processing environments, you don't really solve the problem. You've got IO bottlenecks on one side, if you solve that, you add more functionality, and CPUs are a big part of that too. You add more, you know, I have one person to say, I'm not utilizing my CPU. Well, they want to, so they throw more analytics in, increases the IO bottleneck. So again, this is all kind of in flux, and so it makes sense that let's take it up a notch, okay, high level business value. This is not an SAP question because you kind of laid that at the beginning, but the value proposition for this emerging Hadoop community, you now are offering the big bundle, the bundling packages to help these companies get in and you can work with your accounts, it's not confusing, I got that, check. But for this ecosystem, what is that business value that everyone is striving for? Shoot the arrow forward about a year or two. What has to happen in this ecosystem for the customer value? I think there's three things, right? First off, number one, the developer competence needs to be increased. I mean, everybody's comfortable, you said you did computer science in one of your sort of, 80s, yeah, 80s, yeah, 80s, yeah, 80s, yeah. Old school. I mean, relational theory is great because I mean, you don't need to be a grade A student to know SQL, right? Grade B students, grade C students pick up SQL. It wouldn't necessarily be the case with map and reduce today. I mean, you've got to have a certain skill set. So I think developer proximity to these technologies and having the aptitude to be able to do this stuff. Second thing is, is really there's too much clutter right now in terms of the number of vendors who have just renamed themselves as being become big data companies. We'd almost be able to sort out the wheat from the chaff. Who is truly adding value to the infrastructure and who is just noise? That's why Schnabe, I think, didn't want to put the big data wrapper around there, because I think, at the scale you were running, I think it felt a little cheap, although maybe relevant from a marketing standpoint. Great, well, final question, SAP, what's Sapphire going to be like this year for you guys? Sapphire's going to be noteworthy as usual. There's going to be some good announcements coming out. Tens of thousands of people will of course swarm over across to Madrid this year. We're feeling very good and we've had a great Q3. We're gearing up more towards hopefully a good Q4 as well and ending the year in a high. And we've got some great technology, some great business applications and innovation coming out from the company. Well, we've been very impressed with Bill McDermott and Jim Schnabe because they've been friends of the Q, but we've had them both on for sit-down conversations. Great executives. I will skeptical up first on the co-CEO thing, but I got to say those guys have done a great job. Bill's the sharp-dressed man, very articulate. Schnabe's the mad scientist who puts the strategy together in a good way. I mean, no offense, in a positive way. Crazy scientist in a good way. So congratulations. Thank you. Thanks for coming on theCUBE. Appreciate it to see you. We'll be right back with our next guest. That's SAP here at Strada, integrating in with the emerging crowd. Not playing, they're playing nicely with everybody, doing deals. That's a good sign for the industry and good sign for everyone, including SAP. We'll be right back after the short break.