 Live from the Moscone Convention Center in San Francisco, California, it's theCUBE at Oracle Open World 2014. Brought to you by headline sponsor Cisco Systems with support from NetApp. And now here are your hosts, Stu Miniman and Jeff Frick. Hi, I'm Stu Miniman with wikibond.org. Here in the Cisco booth, Oracle Open World 2014. Beautiful day in San Francisco, but we're here in the booth where you don't know how much it is, how hot the temperature is, except under the lights here a little bit. So at wikibond, we always love talking to the IT practitioners. And one of the first places that you can usually learn about technologies that are being adopted is the IT org inside some of the vendors that are making it technology. So we're going to do a segment here, what often is called Cisco on Cisco. Joining me through this segment is Shashir Kapoor, who's the IT director with Cisco. And Jag Calon, who's an IT architect with Cisco. So gentlemen, thanks so much for joining us. Thank you. Thank you. All right, so we're here at Oracle Open World, and we've talked a lot over those last three days about Cisco UCS. Of course, that really had a ripple in the industry, brought a whole new segment and lots of new opportunity to Cisco and the ecosystem partners. We want to focus in a little bit on how you brought Oracle onto UCS in your own shop inside. So, I want to toss the question out, is somebody give me the kind of the genesis of this project, maybe Shashir, you'll start with that. So, what we did five years ago, we started kind of a broad data center strategy, creating our internal cloud based on Oracle UCS, our product, and also the Nexus, which is. And after creating the broad architecture and strategy, we started migrating some of our critical business applications. So, let me share one of the success stories. Two years ago, we converted, I would say, that time perhaps the largest ERP implementation, our Cisco services, that time was perhaps around 40 terabyte. And we migrated it to our UCS and the TCO, our total cost, reduced by 60%. Very compelling story out there. All right, so, Jag, my previous segment, I was actually talking Red Hat and talking about one of the big use cases that people use was going from Unix-based Oracle to Linux-based on UCS. Can you give us a little bit of a feel as to architecturally, what did you have before? What did you have after? Was it just x86 to x86, or what was the migration? Yeah, what we had before was, we had legacy risk environments, like Shashir was saying, about four years ago. And at this stage, we are 100% UCS. So what Cisco IT has done is, moved a $48 billion business completely onto UCS. If you go into a data center, as you open the doors, you'll see rows upon rows of UCS there. And there is a key aspect of this I want to explain. I actually led this effort, the technical effort. When you're trying to measure a football field, you've got to find the right scale to be able to measure it. Similarly, what we did with this transition was, benchmarking, look at benchmarking. So we'd been doing benchmarking with our legacy vendors, with our legacy products. And we were able to do a similar kind of benchmarking on UCS. And so now we had a one-to-one comparison between the legacy solution that we had and UCS. And we were able to put together all our other environments there. So that created a scale. So this is kind of a call out to other customers that are looking to move in this direction, create that scale within their organization so you understand the transactions per second and the IO throughput that is required for your large applications. So as of now, Cisco's bread and butter, all our large applications run on UCS. All right, so can you help me understand though, was it, were you running the same operating systems and moving that over? Or how many pieces needed to change when you moved over to the new architecture? Right, so the legacy environments, it's a proprietary UNIX environment. So there is a huge cost angle to it. Each of those vertically scaled environments that we had, they were $2 to $3 million each. And there is a huge support cost also that you incur with legacy vendors because only the legacy vendor can support their platform. So with UCS, it simplified that. So that was one of the key reasons why we moved in this direction. We moved over to UCS. And did you add virtualization or is this still on bare metal? Yes, so we have a mix of virtual environments as well as physical environments. Our non-production oracle environments have largely been virtualized across the board, but there are some large environments which are still on physical. So for example, the one that Shishir mentioned, this is the Cisco customer care, the C3 environment. So this is still on physical. So there is, we have, this is around 40 terabytes today. It was 40 terabytes when we did the migration from legacy over to UCS. It is 63 terabytes today and still growing. So I'm curious, Shishir, from a staffing and a skill set standpoint, what did our kind of before and after picture work out? I mean, obviously I guess it's pretty easy to train your hopes on Cisco product, but the size of the team and how much effort it is to maintain and service it before and after. Do you have any metrics on that? Yeah, I can share a few. As I said, specific to this, and in fact, this is the white paper we have published. It is available on Cisco.com and also the Google you can find out. As I said earlier, the RTCO, which is the OPEX and CAPEX part, reduced by 60% compared to this one implementation I'm talking about. Some of the space-wise reduced by, I think, 70%. Well, was that a lot due to virtualization or just the higher performance at a lower density? I would say it was both. It was a higher density and the virtualization and the design of the UCS is much more modular. And our, the electricity, I think, consumption went on by 70 plus percent. Not only that, we increased our total capacity two times and performance-wise also, the bad performance improved by 20%, 30% and so on and so forth. So overall, very stable, very scalable, especially on the database side itself, I had two or three DVS dedicated to this one environment and I have reduced it to one now. So these are some of the success stories. So, Jack, I hear there's a white paper. I'm wondering from an architectural standpoint, did you capture some of the lessons learned and give things back that the field is using or that other IT architects can use based on what you've done? Yes, yes, definitely. It's a very exhaustive white paper. It's around 108 pages. For folks who really want to sit down, get into the details, find out exactly what we did and be able to replicate it. It takes you step by step through the process that we followed from moving from legacy over to UCS. There is a, just to broadly speak about it, right? There are, the model that we followed was, pick up one small environment and cut your teeth, gain your experience on that environment and then be able to cookie-cutter that process across the board. The scaling that I talked about earlier was extremely critical doing the benchmarking, being able to understand comparatively where these environments were with respect to our legacy and UCS, so that really helped us. All right, so one of the tripping points, usually on a lot of these solutions, is the storage. Can you share, is it one platform? Do you work with a few of your partners on this internally and what lessons learned did you have on the storage side? Right, so on the storage side, as you rightly pointed out, there is a shift that is required. Most of the legacy vendors, they will have their own storage solution, right? So don't want to name any vendors, but they have proprietary solutions that they support. When we came over to UCS, we used Oracle's ASM, Automatic Storage Management, and it does require a shift in terms of support. So generally speaking, storage, you rely on the storage team, you rely on the system admins, and you rely on the DBS to do functions in the traditional setup. With the product like ASM, there's more to do for the DBS. So there's a certain amount of shift required, training required to be able to make it possible. In terms of the storage vendors themselves, we use NetApp as well as EMC solutions, and the solution that we used before and after the transition for, as an example, the large environment we talked about was the same storage solution, at the back end. Okay, so Shashir, been a lot of talk about cloud at this show. Cisco, of course, has a large inter-cloud initiative going on. Does UCS help you get ready for, to some of the things like cloud and big data, and maybe you could talk a little bit about how you're thinking about those insights at Cisco IT. Sure, sure, so as I said, we started this journey four or five years ago. We were building, it's called CITIS, which is our internal cloud, and this is completely built on our UCS and Nexus switches. So that's what is the, our building blocks for our internal cloud. Also from the big data perspective, we do have a number of platforms which we are running and they all run on the UCS. So yes, UCS is the building block, and in fact, not only for the, some of the large Oracle service environment, there are other white papers we have also published and calling out. If I can add to what Shashir mentioned for our Hadoop implementation, that we are using Cisco, CPA, common platform architecture. It's the same architecture that is available to external clients. So in a way, we are eating our own dog food. So, Jag, I'm curious when you're testing and working with some of this environment, to some of what you do, feedback to the product development side? Yes, yes, definitely. We work very closely, just as an example, right? The CPA that we deployed for our Hadoop implementation, we are working very closely with the BU to make sure that all the credible feedback is given to them so that they can do product improvements or enhancements. So we have requests for features where we work with them very closely. And I would just also add, even for UCS, we do have a closed-loop discussions with our BU, which is our engineering, and we do influence their roadmap and so some of the features and enhancement which we also work very closely with them. Yeah, so of course, you guys had your Grainslam announcement just a few weeks ago with the latest, there's the UCS Mini, as well as, I'm trying to remember the name of the version that's a highly scalable architecture. If you guys had a chance to get your hands on that, Jack, maybe you can comment as to what you're seeing, maybe if there's anything interesting you can say about the news, yes, units. Right, so we've looked at some of them, we have them in the lab. For example, the B200M4, which has been recently announced, right? I mean, so we are doing some testing on that. We haven't really got the UCS Mini in, but we hope to do it in the new future. All right, so, Shashir, I'm wondering from just the role of IT has been changing. How long have you been at Cisco and, you know, what would? Completing 14 years. 14 years, how does Cisco look at IT internally now compared to, you know, over the last decade plus? Yeah, no, I would say there has been a huge shift. And I think the wild IT runs to make the business running much more efficient. I think that there's a critical role Cisco IT plays as to something we call Cisco on Cisco, which is we are the largest customer of our own internal technologies and gears. I think that's the one thing which we have been doing it. And I think there has been a huge amount of partnership and collaboration going on between the BU's, which is our engineering team, as well as the IT. I would also say the role of the IT is also changing, which we are also working is the way we work. So there's a lot more DevOps model, a lot more kind of what we call it, the continuous delivery. That also we are working with our BU's, also to see if they can also adopt. So I think that's the close partnership going on. So I think I would say the role it has changed. Partnership has increased a lot more and also the way we work. Both things have changed significantly. All right. And Jag, I'm curious what pressures are put on you with things like mobile deployment to your users. How much do you get pressure to change on and make things available through some of the new technologies? Yeah, so definitely I think we want to stay on the cutting edge with new technologies. An example is internet of everything. That's a key initiative within Cisco. And architecture is kind of at the forefront. We lead with architecture. So we are working on a lot of exciting projects that can because with internet of everything, you're talking about a lot of sensor information coming in and that's where big data solutions come in, our Hadoop implementation comes in and we are looking at business analytics solutions to be able to mine this information and be able to make sense out of it, provide more value to our end clients. So I'm wondering, you're going to tell you the internet of everything. When I talked to a lot of users, they understand the idea of sensors, but they have a hard time visualizing some of the specific examples. Do you have anything you could share as to something specifically you guys could do with sensors in your department or working with some of the groups? Yeah, I think if you know Google Glass, right? I mean, if you were to have an IOE glass and you looked inside the data center, you would see thousands of devices and thousands of sensors on them. I mean, a log being logged on the system is also a sensor output. So there's a lot of information that is coming out and a key use case is event correlation. When you're trying to get all these events coming out to be able to correlate, make sense out of them and if a red light goes off in the data center, how do you correlate that to what is the cause of it and what are the different apps that red light is going to go and impact? So that's where all these sensors that we have in the data center, mining that information, bringing all that information together, being able to make sense out of it is extremely critical. Shashir, if I could actually bring it back to the Oracle discussion for a second. When you look at the overall budget for an Oracle database deployment, the licensing's a big piece of it. Is there anything you found in the migration to moving the UCS as to just kind of your IT budget, you know, what kind of before and after you've got a smaller footprint. Over time, did it have any impact on the licensing? Yeah, sir, I would say we do have, of course, with the Oracle is the enterprise license segment, right? So it's very difficult to parse it out is how much is because of this one or two. But as I said earlier, the overall from the license perspective or Linux, our total TCO, which includes OPEX and CAPEX both, we reduce it by overall 60% for that. But one thing I'd like to add, right, like Shashir mentioned before, IT departments are being asked to do more with less. And that is where UCS helped us big time. Like Shashir mentioned, right, 60 to 70% reduction in the cost, right? In the TCO in the cost is extremely significant. It helps us meet those needs that are there. In fact, just add one more point, right? So Cisco IT, we are organized like a service organization. So we have different services. So we all see as a service provider. So I am an owner for database and the ERP platform. And the expectation is each service owner will reduce their run the business costs by a minimum five to 10% every year over a year. So we use certain change the business dollars in order to reduce in a five to 10% per year basis. So that's the expectation for all the service providers as a part of Cisco IT. All right, so run it short on time. I want to give you each a chance to just a final word. Jag, start with you. There's big white paper out there, but if there's one takeaway you can give to kind of your peer architects out there in the industry about moving to a new architecture like a UCS, what would you leave them with? Yeah, I think I talked about the scaling aspect that is extremely critical. So that was the blueprint on which we did everything. Once you have the scaling done, then it is just a matter of the operational teams, implementation teams, just picking that up and running with it. Right, so that's one key aspect. The other is try to do it on a smaller environment. Make sure you, the lessons are learned and you have the standards put out so that it is easy to run with that idea down the road. And also be able to sort of look at the larger picture, plan it for the entire enterprise that you have. And you'll definitely be successful and we're always there to help out in case anything is required. Yeah, so Shashir, I want to give you the final word, but how do you make sure, if you're one of the earliest customers of some of these new deployments, how do you make sure that you don't end up with a lot of the work that might have happened in QA or development standpoint? How do you build that partnership inside the company and any other recommendations you would give to your peers in IT? Yeah, I would just give a couple of quick recommendations. Number one, as Jack said, for UCS, don't think in terms of social service migration. Think in terms of what is your broad data center strategy? Are you building an internal cloud? How is this going to work with the public cloud? How is the hybrid cloud going to work? So think broad, focus on the architecture up front, have a strategy called out, engage with your business partners up front, have your test strategy called out and automate as much as possible all those test cases. I think some of these recipes, we'll call it out part of our white paper as well. And I think that again, the business outcomes speak by itself. Well, Shashir and Jack, really appreciate coming to share your experiences with the Cisco on Cisco, everything going on with the UCS platform, the Oracle migration, good content from practitioners that are doing it in some very large environments. So we will be right back with SiliconANGLE TVs, coverage going to be wrapping up real shortly. So thank you for watching all the coverage here from Oracle. This is Stu Miniman with wikibond.org.