 Live from San Francisco, extracting the signal from the noise, it's theCUBE, covering Oracle OpenWorld 2015 from Studio C, brought to you by Cisco. Now your host, Stu Miniman and Jeff Brick. Hey, welcome back. We are live at Oracle OpenWorld in Moscone, South. There's a lot going on. There's theaters, presentations, demos, all kind of madness. Come on by theCUBE. We'll be here all day. Wall to wall coverage. We're psyched to be here. Joining us next segment by Stu Miniman from Wikibon, esteemed research analyst who we are happy to have on board and Raghu Nambiar from Cisco Distinguished Engineer, Data Center Business Group. Raghu, welcome. Yeah, I'm excited to be here. I mean, 40,000 people in San Francisco this week. I heard 55, 55 is the number I heard yesterday. Yeah, I'm excited to be here. Excited to be in theCUBE again. Absolutely. So we've been seeing you all over. So you've kind of made the circuit like we've had. You've been at Strata Hadoop World, Big Data NYC, Splunk. You're going to IBM Insight. So it's been a busy show season. What are some of the lessons, learnings, things you've come away from? Yeah, let me start with the Splunk conference in Las Vegas a few weeks ago. So we have been a part of it with the Splunk and one of the highlight of Splunk conference was the Splunk 6.3 performance. It was demonstrated on UCS, you know, 50% more performance than a previous generation of Splunk software. Right after that, we were at the Strata Hadoop Conference in New York. We talked about our integrated processors for Big Data. Then last week at SAP TechEd, we talked about Wara, the new highlight from SAP, and the project of Benjamin, a joint collaboration between Cisco and SAP. And this week, Oracle World, and later this week we'll be at IBM Insight conference in Las Vegas, you know. Busy schedule, you know. Cisco's partnering with everybody when it comes to space. You know, yeah, I mean, like we introduced to Cisco UCS five years ago. Tremendous growth, number one market share next year to six blades in five years. It's a significant achievement. 47,000 customers, 85% of all the fortune 500 companies have been invested in UCS. And the $3.5 billion business in five years, you know, it's a lot of excitement. So, Ragu, I think back to five years ago, server virtualization was a real target for UCS. You know, we saw when server virtualization came out, it started buying boxes a little bit different, but UCS had just fundamentally in the architecture, there were things that you did to make sure it really fit the performance, had the right kind of memory. You know, you're looking at things like, you know, Splunk and all these analytics. Can you maybe, you know, talk a little bit about from an architectural standpoint, you know, what do you have to do in the products? What do you have to do configuration? Try from that kind of architecture standpoint. Yeah, I mean, you know, from the very beginning, we focused on two things. One is a programmable infrastructure, not only for like a traditional data management systems, but also for emerging, you know, big data analytics, internet of things. Another one is our focus on a performance. Okay, so far, we have 110 world record benchmarks. If you look at databases or virtualization, you know, on enterprise applications, we are leading the space, meeting our competition, who has been in this business for several years. And this week, we published industries, first DPC XHS benchmark at 100 terabyte scale. DPC XHS benchmark is the first and only, so far, the benchmark for benchmarking big data systems. And we are able to demonstrate, you know, the performance and price performance over the last one year. But today, it is exciting that we are the first event in the republic, you know, at the large scale, 100 terabyte scale. So, benchmarking came up a little bit last night in the keynote. Larry was throwing out, you know, we've got a benchmark, we're throwing it out for other people to benchmark. Explain to people how you guys use benchmarking. Why is benchmarking important? How do you build your models around benchmarking, and how does that directly impact real applications and real customers? Sure, I mean, so if you look at benchmarks in general, there are two consumers. One is internal engineering team. Okay, I mean, whether it is Cisco or Oracle or other companies, they use industry standard benchmarks internally to compare the new product with the previous generation of product, whether it is hardware or software. So once you get good performance with industry standard benchmark, I mean, pretty sure that, you know, at real life, you can expect similar performance improvement. Second thing is the marketing team. How do you showcase the performance and price performance of your product compared to your competition? That is, you know, second consumer of industry standard benchmarks. When you talk about TPC-XX benchmark, you know, there's a performance component as well as a price performance component. And the run rules are specified by the TPC specification. So at the end of the day, the customers are getting a performance and price performance point that can fairly compare with other benchmarks out there. So, you know, when you look at these architectures, one of the feedback we get from practitioners is, you know, when they start with something under their desk and then they want to scale it out, it's tough to do. I know Cisco, Jim McHugh was on talking about this, Cisco validated designs or CVDs. You're doing the same kind of thing for things like Splunk and Hadoop? Yeah, I mean, yeah, Cisco, CVDs, Cisco terms, Cisco validated design. It is a result of a lot of performance and scale testing internally, okay? So basically like if a customer or a partner follows the guidelines in the Cisco validated design, we guarantee out of the box performance and scalability. I mean, typical CVD development takes several months and some of the CVDs in new areas could take even like, you know, couple of quarters before the product is sold. And as I said, we stand behind the performance and scalability of whatever it's articulated in the CVDs. And we have CVDs for, you know, several CVDs for Oracle, okay? From Oracle standalone to Oracle rack with enterprise storage partners. And we have CVDs for the big data for all the major Hadoop distributions and as well as Splunk. So are you seeing the bigger loads? I mean, some loads that were never in a cloud before really start to move and take advantage of some of this horsepower? You know, I mean, cloud is a big part of our portfolio, okay? So from the very beginning, from the web to the world, Cisco has been providing the routing and switching infrastructure for cloud. And there are two technologies I want to highlight. One is ICF, inter-cloud fabric, you know, internet is for connecting devices. Inter-cloud fabric technology is connecting clouds. So maybe like, I mean, you can extend your private cloud deployment to public clouds or you can have deployments across multiple public clouds. And ICF can provide you a seamless interface from an application perspective if it's transparent where the data or application is residing. All right, so please finish. Yeah, I mean, second thing is, you know, private cloud is one thing and we provide our platform for public clouds as well. I mean, there are several of our partners offering Oracle, SAP, you know, and other enterprise applications of service on Cisco-powered clouds. And we have our own cloud, I mean, CIS, Cisco Information Server. And that is another option for our customers, you know, to consider when they are ready for extending their private cloud or extending the private cloud or going full on a public cloud office. All right, could you help us understand how Cisco's inter-cloud fits into this whole discussion and, you know, I know you were on the UCS team, but what interaction did you have with the inter-cloud group? Yeah, I mean, the cloud is, again, a set of tools to extend your private cloud offer to the public cloud. And also, if you have a substitute across multiple public clouds, ICF will do like a transparent access from application level as well as the data tier level. It's so hybrid cloud, right? Everybody likes to talk about hybrid cloud. From your perspective, is it really hybrid cloud or is it hybrid applications that are leveraging different clouds? What's the real reality between hybrid cloud and the enterprise? You know, the application landscape is becoming really, really complex, okay? So every enterprise is expected to have a private cloud as well as public cloud. Hybrid is basically the connectivity, right? I mean, with ICF, one of the unique thing is that you can have a public private cloud offer and based on your application demand, like week before Thanksgiving or week before Christmas, when your system cannot handle the burst traffic, you can seamlessly push the applications on the cloud. From a user perspective, it is transparent. Right, but are there applications that are kind of running concurrently in both instances at the same time or are usually hybrid, people like hybrid cloud is really more the sharing of data between essentially two applications that have some contingency. Some of these technologies are emerging, okay? I mean, in some of the data management systems, you can have an instance in your private cloud and another instance in your public cloud and do load balancing, okay? And the second one is you have instance either here or on the cloud, but the data is shared between these applications. Right, right. So this is like, this is emerging, okay? I mean, and we believe that hybrid cloud is going to be something of the future because every enterprise will have, you know, deployments in their own data center and subscription to one or more than one public cloud office. Yeah, Raku, you bring up a real good trend that we're seeing a lot is customers want to be able to access not only their own data, but either there's community data or their data sources that they have. And heck, you know, we haven't even talked about IoT yet because then, you know, we're just going to have data everywhere and, you know, what's at the edge? What do I pull in? How do you see kind of this dispersion of data? And of course Cisco, you know, networks many of these pieces together. How does that get into architecture? You know, internet is IoT or IOE, internet of everything is a big part of our portfolio. I mean, one interesting observation is that if you look at the analytics today, 80 to 90% of the analytics are happening in your data center. Okay, that's going to change. By 2020, the prediction is that, you know, 80% of the analytics are going to happen at the edge. Okay, like in certain cases, you know, the curvy will be pushed to the edge and processed in a distributed manner and only the result will be sent back to your application. I mean, this is going to be a game changer, the whole IoT-based applications. When you say, when you talk about pushing to the edge, pushing the compute out to the edge, because it's kind of both sides of the coin, right? Because more and more sensors, smaller, less power, more ubiquitous, those won't have the necessary the capability to do heavy lifting, computing, or heavy lifting, the data aggregation. Other things, you'll be able to push the compute out to the edge with the data. So it's just, it's still this constant, good for Cisco, moving data and moving compute all around. So we will have, I mean, we already have and we will continue to innovate on, you know, edge devices, whether it's a router or a server for the edge, having like more and more intelligence so that it can process data locally before it is sent to, you know, in your data center. Yeah, so it's crazy. I mean, Internet of Things is going to be ridiculous. We're already seeing it, of course. Everyone, the biggest Internet of Things that some people like to say is they're Tesla. That's right, they're the world's biggest wearable. So what's coming up next? The pace of change in this industry is so ridiculous. You just said you've been to all these shows, as Merv Adrian likes to say from Gardner, who was at the Big Data NYC. You know, it's data from data and rest of data in motion. We're seeing Spark, the emergence of Apache Spark. So what are you kind of excited about for the next six months? I do a lot of exciting things, okay? I mean, I don't know which one is not exciting in this world, okay? So, I mean, Spark is, of course, one of the interesting technology, the ability to run things in memory. We will be announcing a set of reference architecture for Spark in the near future. You know, another thing is all the emerging technologies on the storage space, like SSDs, are becoming pretty common in terms of capacity and performance, of course. And the next generation, NVMe. I mean, you know, that's going to be a game changer. With SSDs and other technologies, it is like bringing a storage closer to the compute. With NVMe, it is pretty much bringing storage inside the compute. You will not only have the performance, but also have the capacity. If you look at our B200 blades, that is one of the most popular computer systems that we have in NVMe is reality. We will be able to support, you know, 16, 32, 64 terabyte of storage capacity inside one single blade. So, Fragu, I was seeing in your bio that you participate in some standards work. And one of the things we've been looking at is kind of the open source initiatives, everything. Linux Foundation, we talked about Apache Spark. You know, how do you see, you know, we see it changing somewhat, you know. Standards, you know, open source, that obviously there's an interplay there. How have you been seeing that changing, and you know, what do you see the role of standards going forward? Yeah, I mean, standards, like my primary focus has been industry standards for benchmarking systems. So if you look at the TPCXH's benchmark standard, I was the chairman of the standards committee. The benchmark was a joint effort between several companies. If you look at around, like most of these companies are part of the TPCXH's benchmark of standards development. Recently, I got elected to chair a new standards committee for IoT to develop a set of standards for the benchmarking internet of things. So that way, the various products and technology can be compared in terms of performance and price performance, you know, fairly. Open source, you know, technologies are pretty interesting. And Cisco is a big supporter. Has been a big supporter for open source technologies. I think it's a power off of the community, right? I mean, we have seen that, you know, in Hadoop and other major software distributions, the speed in which the feature sets are implemented in the open source community versus the traditional solutions. I have a lot of respect for traditional solutions, but with respect to speed, the community effort, nothing can beat that. Great, well, Ragu, thanks for stopping by. We'll let you get back to your busy conference schedule. Give you the last word before we go out. What are you most excited, what have you seen here at Oracle that surprised you? Anything? Well, I think, you know, two things. One is, you know, all about utility computing and the digital transformation. We talked about that, you know, last time. Right. A lot of exciting things happening. Good, all right. Well, thanks for stopping by, safe travels. I'm Jeff Frick. We're live at Oracle Open World. You're watching theCUBE. Stop by. We're on the exhibit floor. We'll be here three days wall-to-wall coverage. We'll be back on our next guest after this short break. Thanks.