 The Cube at EMC World 2014 is brought to you by EMC. Redefine VCE, innovating the world's first converged infrastructure solution for private cloud computing. Brocade, say goodbye to the status quo and hello to Brocade. Welcome back everybody, this is Dave Vellante with Jeff Frick. Daryl Smith is here, he's the chief database architect for EMC Global IT, EMC's a $25 billion company and soon to be a $25 billion company, quite diverse, very demanding, global, now the Federation plays into that. So Daryl, welcome back to the Cube, it's good to see you again. Thank you very much, appreciate being here. So how has the business evolved? I mean the Federation now is a new sort of dimension of the EMC business, how has that affected you? Well it's provided some more opportunities in terms of different technologies that are now available to us, basically bringing in the best of breed of the different parts of the company and putting them together to build really platforms that allow us to fully automate our entire infrastructure step. You mentioned best of breed, as an IT practitioner, now you work for a technology company and a technology company is going to say you're going to use our stuff, your own dog food or drink your own champagne as they often say. But the age old discussion in IT has been do I go best of breed or integrated suite? Best of breed, I get the latest and greatest technology, I can drive competitive advantage, integrated suite, I might be simplifying things a little bit. Is that sort of spectrum still valid or have you been able to achieve a balance between best of breed and integrated suite? We really go for the balance. So I mean you can buy a suite that advertises that it has everything that you need and it's very simple and it's click and drag and drop. You can get it all done but what you're going to find is that suite is going to assume that you're going to do things a certain way and there's potentially a better product or capability out there that you can integrate with that suite to really get you that final level that you need to get to. So talk a little bit about, I mean you guys are a big Oracle shop, you're an expert in that field. Talk about what you've been doing with database and database as a service. What's new there? Yeah well that's really exciting. For me the biggest part or the biggest pain really in terms of managing databases is getting them deployed. So it really takes efforts from storage, from systems, from backup recovery guys, from now virtual administrators, DBAs all working with a lot of different touch points to get that product out the door. Well what we've done is we've basically removed all of those middlemen and replaced them with automation. We didn't actually get rid of the people they still work in but in terms of getting the database out there we now allow the DBA himself or herself to order that database from a webpage, configure it with whatever number of CPU, memory and storage it needs and then literally automation will go build it for. So in the end of the day I can have that database up and running in under an hour. So give us a drill in those metrics a little bit. What was life like beforehand when you had to get the storage guys, the systems guys, the backup guys, the database guys, the virtual guys? What was it, how long did it take you to deploy then and now you're saying it's under an hour? What was it before then? So literally it was a two to four week process. Which actually is pretty good. Well thanks to virtualization that got us out of that four month process which we've talked about before. But still even though it's virtualized literally you could go spend up a VM in 15 minutes there's still all that process and procedure and policy generating tickets, trying to get someone to work on your ticket, that really takes time. So let's take a specific example of policy and how you've automated that. So I think of protecting the data. Sure. How does the policy get automated? What do you do? You set up sort of a spectrum of protection capabilities and you sort of dial it up to the one you want. I mean it's not an infinite spectrum I presume. I presume it's a series of buckets that you can choose from. Oh we can take that one example of the backup. I'm going to back up daily, weekly, incremental, full. How does an IT pro interface with the system? Is it high, medium, low? Is it specifics around the particular RTO, RPO? I wonder if you could take us through an example. Sure. So that's a particularly painful process typically and has been for us for a long time where backups aren't really fully in the control of the DBA. So I've got to go install an agent. I have to get that agent and that client associated with a particular storage or a backup system put into a storage pool in order to back up to a location. Then you go to another group, get that scheduled so they can run daily or weekly and then I've got to script the backup itself. Very painful, high touch process. We've automated that also. So now from the webpage one of the options is do I do a daily backup? Do I do a backup every other day? Or just do a weekly backup? And I can just select that and the automation goes and does all those other things that were touch points. How does the policy get determined? Is that an IT business liaison or is that the business direct input? How do you decide? So it's all based off of the RTOs and the RPOs. So if my recovery time has got to be an hour, well that's actually something even more than just a typical backup. That's going to require me to take snapshots or clones periodically through the day to get larger databases into that RTO. If it's something where the RTO can be four hours or I can do a standard daily backup and give me that same recovery time even for large databases. But if it's not quite that high maybe it's a business supporting business critical or a test dev, then my RTOs are much larger and I can go for a weekly backup or every other day type backup. It's really up to the DVA to understand what I can get based off the size of the database and the restore times. Which again, by putting the DVA in full control of the backup and full control of the recovery, now that makes it more of a consistent backup and a consistent restore because you can manage it. It's no longer create a ticket home but somebody starts working on it and you got to try to figure that into your RTO. The DVA is in full control. So Daryl, now you can spin these up things way, way easier. If you build it, are they coming? I mean, what's the impact on the business now that you can so much more easily create this infrastructure for them to execute on? Do they know? How do you communicate that? Are they seeing it? Are they spinning up more things? What's the impact of your customers? So really the goal for this database as a service platform that we built was to give the DVAs the control that they need. So what that's allowing us to do is respond much more quickly to the business demands. Take for example, it's end of quarter and the database is running poorly because it doesn't have enough CPU cores. In the past, that'd be, hey, that's too bad, we'll get to that when we can. Now the DVA can literally just go in there, add more cores and it'll automatically add it online in minutes. It's really as far as the business concern, it's allowing us to be so much more agile that the business now believes that we can do things and we can do them quickly. And they're not looking to try to maybe source around us, work out to the cloud providers that will offer that capability. So it really just allows us to compete more and better for that business than we could ever do in the past. How about, let's come back to the automation. So how did you automate all that stuff? That was kind of a trivial. I mean, you obviously had to understand and document that, but then how did you go from that mess to who fully automated, what tools did you use and what was that process like? How long did it take? So getting back to that, you know, by the suite or best of breed technologies, we used VMware's vCloud Automation Center as really the foundation for that IT as a service capability. So that gave us the portal, that gave us the ability to generate VMs, virtual machines very, very quickly with the OS pre-installed. We then augment that with Puppet from Puppet Labs, which gives us a manifest that installs the database software, configures the OS, adds the additional storage or I'm section VCO, basically puts that machine into the final state it's supposed to be. It's Puppet Labs, the state enforcement application. So I take, you know, best of breed technologies, vCloud Automation Center comes with VCO, it lets us do a lot of the infrastructure provisioning and couple that with Puppet to give us that full state management to get that done. And what was the sort of, that was a project in and of itself. How long did that take? So we had to start from scratch. So basically for us, we're looking at about a two month project to build that all out. We had done a POC earlier, so we kind of knew what we did. We kind of worked out all of the design criteria. So, you know, we went in fully designed and we went in, you know, coding basically. So for us, it was about a two month process. All of that code, all those scripts, we've actually made available to our consulting organization. So their goal is really to go out there and be able to do that in, you know, six, eight weeks to get somebody up and running. And typically this kind of a project is for Greenfield. It's for new databases, especially when it comes to the day two stuff, like adding more CPU, adding memory, or adding more storage, which is really where DBA spends most of its time, trying to get that kind of work done. We're ingesting all of our pre-existing databases, right? Thank goodness we're, you know, 86% virtualized in the database space. And that works with Oracle, it works with SQL Server, it works with Postgres. We're ingesting all of those existing databases so that we can provide all that day two functionality to those databases as well. Now do you have, so how do you keep track of all this stuff? Do you have an IT service management system, or does that all work? So there's a change management system, right? A CMDB change management database, which we keep in sync as things change. So if I added more CPU, immediately the CMDB would get updated. There's also included in this, is our own development agile process. So every little one of these pieces, all these different automations, you know, require different modules, different workflows, different blueprints. And trying to get those all to work together really requires a team effort. And to manage that team effort, you know, basically we used a scrum process within the agile framework. And all of our code gets checked in and it's constantly being tested. So talk about the, we had Nancy, major's on just now from Brown University. She was talking about the sort of organizational tension between the DBAs. Yeah, he gave a little heat to the DBAs. Still her storage admin, yeah. Well, I think, well, she said, I think it's perfectly reasonable what they were asking for, we just couldn't deliver it. Right. And it's true. Well, it's the black box, right? That you were talking about. And it's true, DBAs are kind of persnickety about their data, you know? It's kind of the way you want it if you're running the company. Right. Well, you're responsible for that. You know, ultimately, if there's a problem, even if it's a problem with the application, everybody goes to the DBA. It says, what's the problem? Right, because the DBA has a full insight into what's going on with the data, the actual data, what's going on with the data platform, as well as how's the application interacting with it. So the DBA really has the ability to manage that environment. And so you need to give him that kind of control and that kind of insight, right? With this full automation, we can provide a single pane of glass for monitoring using VC operations manager. So that I can see storage metrics, I can see virtual metrics, the ESX metrics, the VM, the OS, the database, I can see it all put together in one place. So let's give the DBA the ability to manage the platform that he's ultimately responsible for. Now, you mentioned before, that as soon as you automated this process, you had all these people involved. You said you didn't get rid of the people. What'd you do with the people? Did you retrain them? Did you redeploy them? How did that all go? I mean, redefined them. Redefined them. So basically, a database administrator, his job is to manage that database platform. That includes doing all these installation tasks, adding data files, getting other people from other groups to do some of the work for them. That's very time consuming. So what we've basically done is removed all that time consumption and it allows the DBA to focus more on being proactive, focus more on making sure that the databases are running smoothly, working with our business units to make sure that we're providing them the level of service that they need, making sure that our backups are valid, maybe even testing restores once in a while. It allows us a lot more time to be able to improve not only the platform, the database platform, but also ourselves. And so our DBAs now are learning a lot more about networking, a lot more about virtualization, about storage, to really give them a better understanding of how the database works. Testing restores, that's a novel concept. Yes, we always do that. Oh, everybody does. Okay, well, this is Darrell. Thanks for coming to theCUBE. I want to ask you the last question is EMC world. How do you spend your time there? You know, a lot of people, a lot of IT practitioners say, I meet my peers, I talk to them, exchange ideas. How do you spend your time? Are you getting dragged into customer calls? Do you have time to exchange ideas with your peers? You're coming to theCUBE, which is great. Yeah, so I've definitely had a few customer meetings. I'm here. And to me, that's really the fun part of coming to EMC world, really getting to talk with other people, about the things that really energize me. But at the same time, there's a tremendous platform here to learn. There's not a lot of my peers here. There's not a lot of DBAs. I think as the years go on, we're going to see a lot more of them. Because remember, I freed myself up from having to do all this day-to-day stuff. I'm off learning about best ways to migrate databases from BMACs or RACs. I'm learning about best ways to manage your data lake, as data explodes. And so there's really a great learning opportunity for me to grow myself. Okay, and so this is sort of a very focused show for you, right? Obviously, storage focused. And then you attend Oracle Open World, of course. That's really where your peeps are hanging out, right? That's where my peeps are at. All right, Darryl. Thanks very much for coming back on theCUBE. Really appreciate your true IT pro. And it's always a pleasure to have you on. Keep it right there, everybody. We'll be right back. This is theCUBE, we're live from EMC world 2014. And we'll be right back.