 Okay, we're here, it's live, day two, Oracle Open World. Moscone is draped in red, that means the Cube, and this is the backup and recovery spotlight, focus on data protection as a service. It's a theme, John, that you and I have talked about since, really, we started this back at VMworld in 2010, this notion that backup has to transform, it has to change. Daryl Smith is here, he's the Chief Database Architect within EMC IT, we love to talk to the practitioners within EMC's IT division, very leading edge, set of folks, Daryl, welcome back to the Cube, good to see you again. Thank you very much, appreciate it. So this is kind of your world here, Oracle Open World, all the DBAs hanging out, the whole 60,000 DBAs, I guess, well, maybe tens of thousands of DBAs. So how's the show going for you? This is great, I'm really in my element. This is the place I want to be. Yeah, so what have you been doing since you've been here? Just kicking the tires, they have you working, meeting with customers, are you able to check out what some of your peers are doing? Well, it's been great, I have met with several customers, obviously I've been out here in the exhibit hall floor, there's a lot of great technologies here, but primarily I'm here to get to learn, I'm going to sessions, finding about where the trends are, what's going on with 12C, a lot of great stuff. So what do you make at 12C? It's interesting, there's some really nice features in there, some features I have yet to hear about, and then there's other features that don't really make a whole lot of sense, I think in some cases. Yeah, like what? You're going to nail me, aren't you? No, no, what makes sense, what's confusing to you? I mean, tell Oracle, maybe try to fix it, but no, so let's start with what do you like? I mean, from a practitioner's perspective, what about 12C makes you say, oh wow, I really want that, thank you, Oracle, for doing that? Right, so there's a lot of features that really help with the storage. They've made some great improvements in the prefetch, table prefetches and things like that, very good. There's a lot more integration between the different product sets that they've got, some of the active data guard improvements are really phenomenal. Some of the Golden Gate improvements have been really phenomenal, so there's a lot of great technologies that are packing into it. Right, okay, so the big part of that too is in memory, right, and you guys have some experience with that, I mean. Right. I mean, in memory's been around, that talk's been around, it was, the database has been around. No, no, times 10's been around for quite a while. Yeah, right, so, okay, so that's appealing to you. So you see you guys taking advantage of that over time, is that right, or? Well, you know, it's obviously something we're going to have to look at. There's a lot of in-memory database players now. You know, SAP has HANA, Pivotal has SQLFire, we're definitely evaluating those. We're already starting to use some HANA, some SQLFire. So we're going to have to take a look and see how the implementation was done with 12C, and if it fits our workloads, we'll start using it. So talk about what gets you excited as a DBA. We always hear about doing more with less, you have to do more with less as well. I mean, it's the mantra in IT, but you know, wake up in the morning and say, oh, I got to do more with less. You're probably thinking about, all right, you want to work, how am I going to add value for the business, how am I going to make the line of business more productive, how am I going to drive value, right? So talk about that a little bit in your role as a DBA. Yeah, this is really a very exciting time for me. I've grown up with databases, I've been doing it since Oracle version four. So I've been doing that for a long time, but this next wave of computing is really all about automation, standardization and automation. So the whole IT as a service mantra requires that. I need to be able to automate things so that my DBAs aren't spending their days adding data files and monitoring backups and checking to make sure that the system is running well. I need to automate all that so we can then focus on really getting to that front office, getting in front of the business and understanding what their true problems are so that we can be much more agile in response to those. So I don't know if you were able to hear Stephen Manley's little story about he was talking to a DBA at one point. And the DBA basically said- He kind of threw a son of a bus. You said it again? I said, I think he kind of threw a son of a bus. No, he didn't really throw the DBA in the bus, he kind of threw the backup admin under the bus. Right, well, good point. I guess he did, I mean, he's not talking about you, Darrell, I know you a little bit, you're a very respectful guy, you would never throw your storage admin under the bus. Not live anyway. But what he said was that the DBA was frustrated because he or she didn't have any visibility on how the data was being protected and the backup admin said, trust me, I got you covered. Okay, but I want to know more about it. So talk about that a little bit. Yeah, so talk about that a little bit. What was, what did you want that you couldn't get and then we'll talk about how things have changed. Sure, so DBAs hate black holes or black boxes. That's definitely a theme that I heard from that last interview. We like to know what's going on, we need to be able to get our hands on it and touch it and feel it. And backups are one of those things where you kick off your arman backup, maybe it's going to complete, it's running slow, you really don't know why. And then you go talk to the backup admin or you create your ticket and they're like, everything's fine, don't worry about it, it's covered, but the backup is now taking five hours and six hours and you really don't know what's going on. So how can you really trust that it's safe? Nowadays, things are much simpler. Take for example, Enterprise Manager, not OEM, but the Data Domain Enterprise Manager, I can actually go in and see if the backup is running and it's running at peak. So I can truly estimate when that backup is going to be done. Darrell, Jeremy Burton was up on stage in the keynote after Joe Tucci gave the awesome intro, but he's getting into the weeds and getting into the product details and he's talking about the DBA and David and I were just commenting on the intro about the balance between the DBA and other things like recovery, real critical managing the data loss side of it. But Jeremy kind of points out that the DBA roles obviously change with automation changes the game a bit. I mean, it's not replacing the DBA, all that talk is kind of a little bit over the top, but in general, the DBA role is becoming much more versatile, much more needed skills in other areas, whether it's more of analyst role, data science, or getting more geeky into the app side. So in terms of technical, you can be more analyst on data science or you can go in and go deep in the weeds on technology and get into the app side, coding. How do you see that balance and that transition of the new DBA, the modern DBA, where there's range now of skill sets and functionality that they have to do in their job? Right, so referring back to the keynote, it's both disruptive and opportunistic. It's a great opportunities for DBAs to really get themselves out of the weeds. Does everybody really want to add data files and go monitor and check to make sure the backup completed and do installs of the same thing day after day? We don't really want to do that. We want to really watch ourselves. Give me a gun, I'll put it in my head right now. Give me the time to get out there and really understand what our customers need so that I can respond to that. And that's going to not only help me grow personally, but also professionally. So it's a great opportunity. What's some of the vibe out there and the trenches with other DBAs and the peer group out there and the practitioners? I mean, what's the sentiment out there for DBA? Do they like chomping at the bit to do more things? What are you seeing in terms of just general anecdotal trends that you're seeing? Well, once we get past the, you know, I'm drowning and I can't get up, talk in conversation, because we really don't have the time because we're so stuck in the weeds dealing with this growing amount of data. You know, we need some help. And so once we get past that conversation, we start talking about how automation and service-oriented architectures can really help get you out of those weeds. It's a very positive conversation. It's liberation, really. I mean, the DBA gets liberated. Hey, thank you for pulling me out from a certain death. I'm turning the death of boredom, I think. Yeah, the drowning analogy is a good one. I mean, I think the lifeline of automation, DevOps, who's seen with, you know, agile programming has kind of really hit the mark and made the data component as a service very relevant. And I think, you know, EMC and others are talking about this data fabric as an enabler layer. That's not an admin function. It's really more of a robust business mission critical opportunity. It is, it absolutely is. I want to talk about IT transformation, data protection as a service. I want to test some things out. So, you know, you hear the vendor messaging around IT transformation, not just, you know, EMC, but other suppliers. When you talk to practitioners, we like to test that. So where are you guys, in terms of getting to so-called IT as a service? We've had Sanjay Merchandani before on theCUBE and your new CIO and that was always a vision set forth. Where are you in that so-called journey in terms of getting to IT as a service? And then I want to talk specifically about data protection as a service. Sure, so we've been building IT as a service for a while now. We've built infrastructure as a service, which is interesting. We've built databases as a service, which is interesting. So building all these little components out, you know, they're helpful. But really where we're at now is the full automation. So we want to automate and we are automating the entire application stack right from the load balancer all the way down through the database with built-in data protection backups, built-in security, built-in DevOps model, right, with the SDLC integration. We want to go for full automation so that we can really raise all of our common practitioners, our DBAs, our CIS admins up a level. And what's the goal there that you literally have no human intervention other than swapping out a disk drive when it breaks kind of thing? Well, that would be great. I don't think we're going to quite get there. Maybe, you know, when we get to the age of Hal, right, for those 2010 enthusiasts, we might get something like that. But reality is I don't need to spend a lot of hours deploying applications. I don't need a lot of hours configuring systems. I really should spend my time focusing on how can I make that Oracle database run to perfection? And so take it up from, you know, stuck down in the weeds, I don't have time to see what's actually going on and just bring our whole skillset up a level. So that we're focused on the internals versus the infrastructure. Okay, and so if what I described is Nirvana and we probably not going to get there in our lifetime, but, and how far can you take that automation? You're saying provisioning, you said load balancing, and you're saying tuning still a lot of manual intervention, obviously. Tuning the application is going to require human intervention. And that's where you actually want to spend your time. Is that right? Okay. Why not be able to provide a better service for the business versus just trying to survive? And now I want to talk about data protection as a service because in a way, data protection used to be pretty simple. So it was, okay, one size fits all. Here's what you get. Databases were like 500 meg back then. Right. Of course, you know, the bad thing about that was that you either spending too much or too little on protecting your data. So where are you now with regard to being able to say, okay, this application based on its business value, its RPO and its RTO or whatever business impact analysis you do should have this level of data protection and I want to spend this much versus this app. I want to dial it down or dial it up. Can you actually do that today and how do you do that? You can. I mean, you can literally set service levels for backups. You could set priorities that, you know, one more mission critical system gets the backup resources over another. But I like to think about it a little differently. If I can go ahead and I can automate it all and I've got automated provisioning of my backups. Right. It's very easy to do these days with tools like VCO. And because I've got technologies like data domain boost and data domain itself and RMAN and all these integrations that happen, it's very easy to drop down the DD Boost agent and fully automate that backup. So now that the DVAs have no issues in terms of doing the backups, now throughput has been almost eliminated. Because I'm now doing my D-Doup on the server, essentially I've decentralized that function. So now my backup media doesn't have to worry about doing all that work. I can actually treat every database almost identically and not worry about any one of them losing any of their SLA. All right, so let's drill down to that a little bit. So RMAN is obviously, you know, the oracles. It's really the only backup methodology. If you're going to back up or actually you really should be using RMAN. It really is the only choice. So okay, so, and that's the prism through which you look as a DBA through your backups. It's RMAN. So, you know, everything else is sort of secondary. You want it to integrate with RMAN. So you say data domain has done that and others have done that as well. And now talk about how where DDBoost fits in all that. How do you leverage DDBoost to create a business capability? Right, so to me, DDBoost is really the secret sauce in terms of making this all possible. So DDBoost is integrated with RMAN. So I can use RMAN and just literally straight back up to the data domain appliance. And then I can also have RMAN instruct the data domain appliance to make a copy into a disaster recovery type site. Right, which, you know, we use for, we're DevRefresh right from production and also for offsite tapes. Not really tapes. Oh, you use tapes? I'm sorry. If I say tapes, I'm supposed to say that other word. Can we say that? Yeah, you can say tape sucks. Okay, tape sucks. Okay, so you use the name tapes because that's the nomenclature of the syntax? That's, you know, I'm old school. What can I say? Or do you actually use... We are tape free. EMC uses no tape. We use no tape. Really, not even for deep archive compliance. When you're de-duping at that kind of a level, all I'm tracking are my changes. So the amount of physical storage required to do database backups is very, very small. Add to that the fact that I've got built-in compression. I literally backed up a 12 terabyte database. First time, no de-dupe down to three terabytes just from compression, which is amazing. So now my next day backup is literally going to compress down to another 100 meg. We can store an awful lot of data on that data domain appliance. And if it starts filling up, we can actually archive it using data domain. And you've got remote sites that you can fail over to in the event of a disaster. Absolutely. That's kind of your approach, as opposed to shipping a bunch of tapes on a truck and driving it to a data center. That's not something that you guys are doing. No, tape sucks. Okay, but if you didn't work for Joe Tucci, would you use tape? Not if I could help it. Okay. As far as I'm concerned, tape should be dead. The technologies are there today. Interesting. Okay, good. I didn't realize you guys were 100% tape-free. I thought you did like the last resort. There was a time where we would back it up to data domain, and then after a while we would drop it to tape and take it off-site. But now that our data centers are 600 miles apart, I can literally back up to one data domain appliance, have it replicate to another, and that takes care of all of my regulatory needs in terms of the off-site storage. And because it de-dupes down so small, it doesn't really require me a whole lot of this space. Data domain has good tape integration for those customers that don't want to do tape. It does. As you will know. You just hate tape. I did not. I did not. All right, Terrell, I'll give you the last word here. Advice for practitioners trying to get to IT as a service generally, but specifically data protection as a service. What's your advice? What are your roadmaps? What's your prescription for success? Well, I think the key is two-fold. One, we want to make sure that the DBAs are in control of their backups and their recoveries. Most importantly, the recoveries. This is one of our biggest problems. We need consistent backup times. We need consistent recoveries. Using a tool like DDBoost, going straight to data domain, whether that be through a traditional backup media like Networker, or using OEM to actually manage that backup, puts the DBA in control without having to get the backup admin involved to try to get a restore done. But it also gives the backup admin the full control to manage the capacity. Excellent. All right, Terrell. Listen, thanks very much for coming back in the queue. Appreciate you sharing your perspectives and have a good rest of the event. Thank you very much. Okay, this is Dave Vellante with John Furrier. We'll be right back with Caitlyn Fordham. We're going to talk about the portfolio that EMC has, how they're applying it to Oracle. We'll be right back after this.