 Live from the Fiat Barcelona Grand Via Compensator in Barcelona, Spain, it's The Cube at HP Discover Barcelona 2014 brought to you by headline sponsor HP. Barcelona everybody, this is The Cube, my name is Dave Vellante and we're here at HP Discover 2014 in Barcelona. Patrick Osborne is here, Patrick is a Cube alum, runs the HP backup product business. Patrick, great to see you again, and Frederic Chalal, great to see you, thank you very much for coming on. Frederic is the head of infrastructure and engineering, division at OECD, so welcome. Thank you, it's a pleasure to be with you today. So, why don't we start with OECD, what's the organization all about and what's your role there? So, the OECD is an international organization, multi-governmental organization based in Paris with 34 member countries. So, we have member countries in almost every continent, like the US and Canada and North America, we have Chile, South America, most of the western countries and also Australia, New Zealand, Japan and South Korea on the east side. So, it's a worldwide organization. And your role is head of infrastructure? So, I'm the head of the infrastructure division in the IT department, the infrastructure division is looking after server storage, virtualization, network security, everything which is not, I would say, business related, so the foundation of the information system. So, not applications? Not, we are managing a few applications, but it's what we are considering core applications like the email system, the databases, collaboration software, but not products that we are developing for our users. Yeah, so infrastructure, software, middleware, database, tools, email, you count as infrastructure and everything on down. Okay, big job. I always like to ask practitioners like yourself, what's the business driver, what's the, where the pressure is coming from? One obvious for you is the global nature of your organization. Maybe you could talk about the drivers in your business. First of all, the OECD's mission is to promote and recommend economic policies on various subjects. It can go from taxes to health to technology to tractors or fisheries. I mean, we are covering almost every economic domain. And we are not only the 2,500 people working at the OECD are providing this recommendation, the OECD is also a place, a forum where the governments, it's almost 40,000 international experts that come to parents to the OECD to discuss about their problems and they try to share experience and find common solutions to the global problems. So we are working a lot with governments, but not only governments, also central banks, international organizations. So we have 34 member countries, but we are collaborating with almost 120 countries around the world. So information access, collaboration, data protection, all big challenges for the OECD. So we have different areas of expertise. Of course, we have economists working at OECD. We have a lot of statisticians who are trying to aggregate data that are coming from the National Statistical Officers and try to compare the numbers and provide a unique set of numbers that can help economies to build the new policies. So one of the main challenges that we are facing and we have been facing this challenge for a while now is the data growth, data explosion. So we have approximately 42% data increase every year, which is of course a big challenge in terms of capacity management, but not only capacity management, it's also a challenge in terms of making sure that the data is secured, it's accessed on a permanent basis, it's backed up in case we have a problem and we need to restore it. So one of the challenges that we have been facing and trying to improve on a daily basis. Alright Patrick, let's turn to you. So not an uncommon problem that we're hearing from Frederick. High data growth, most of that data I'm sure is unstructured. You've been at the files, I had a thanks for a while now. You're hitting your groove swing. It's been several years now in this job and your portfolio is starting to round out. So tell us where we're at at Discover. You guys got a bunch of new product announcements in your area. You did the unveiling last night. Always love the good unveiling. Good stuff. So bring the audience up to date on what you guys have announced here and where we're at with the products. Yeah, so like we do every Discover, we jam packet full of innovation, right? That's what we're here for. That's what HP is famous for. And this, you know, go around was no different. We had a large number of announcements on the primary storage side, which I'm sure you're going to talk to David about later. In terms of data protection, we have some solutions for three part, right? Hottest, fastest growing platform from primary storage in the industry. And we get a lot of customers who say, what's the best backup solution for three part, right? So we decided to go solve that problem and then go even farther, right? So giving the ability to federate primary storage with secondary storage, so the store wants recovery manager central, product suite that we introduced, we announced yesterday. We talk a lot about simplicity in backup. So, you know, no one likes backup, right? It's more of a TCO model, right? And we try to make it as scalable, as simple and as efficient as possible. So we really double down on store ones, federated catalysts, a large number of expansion in terms of ISV support and different solution areas. And then, you know, we, to address some of the unstructured data needs, we came out in two areas with store all, integrating with our control point software, gives the ability to take a look at big unstructured data repositories, catalog that application data, do a lot of metadata management, then you can affect policy on it. And we actually announced a tech preview for some very powerful technology in the areas of object storage that we're going to be releasing in 2015. So a lot of, a lot of innovation going on. A lot going on, right? So, and these are all new products, right, as of this week and shipping when? In January? Yeah, everything. Yeah, all available in January. On top of all the great new innovations we have on top of 3PAR and in our hyper-converged system. So, a lot going on in HP storage. Okay, Fredrik, let me come back to you, just to sort of describe your infrastructure and some greater detail, generally, but specifically, let's get into the backup piece. Okay, so in terms of infrastructure, we are an organization based in Paris, so we have in Paris two data centers. In terms of hardware, we are almost relying only on HP hardware, so we are using HP blade systems with blades to us, all our virtualization environment. We are using also HP storage, so we are using HP 3PAR, but also older EVA storage arrays. And to backup this environment we introduced two years ago, we stole once 6,200 appliances, to help us to deal with this huge amount of data that we have. How much data? We are talking today about approximately 70 terabytes of data, which is not a huge number compared to other organizations that I was able to meet in Barcelona this week, but we have a real growth in terms of storage, because I was mentioning to Patrick earlier, in 97, 17 years ago, we had to deal with only three gigabytes of data, and 17 years later we have 200 times more data than we had in the past. Not only we have a slight increase in German servers and storage, but in terms of data, the challenge is to be able to deal with this amount of data on day-to-day basis from a backup point of view. Well, you'll be over 100 terabytes shortly, so that's... Probably shortly. That's why when we implemented the current storage solution, we had this 200 terabytes on the horizon, and we tried to implement a scalable solution so that we won't have to come back to our senior managers to ask for more money to be able to backup and backup the data. So how long have you been with the organization for a number of years? I've been working for 20 years. Okay, so you have the history of the before and after, so I wonder if you could... you mentioned you brought in the 6,200. What was... how were you protecting data before that? What was the backup journey, if you will? So initially, and I said we were backing up a small amount of data. It was huge 15 years ago, but in the end, it was not a lot, and we were using standard tool of the market with tapes, essentially, we were backing up tape, and backing up tape is the real, I would say nightmare now, I can say, because the tape is not reliable, it takes a lot of time, the stores are very difficult. So in 2009, we went from a pure tape backup environment to backup to disk, this disk-based backup with already HP technology, but we made a lot of mistakes in the implementation. We were looking only at the technology aspect without looking in much detail about scaleability, about cost, about security, about high availability, and three years later in 2012, we had to do a complete review of the backup infrastructure because we had really different requirements and objectives than three years ago. So the main requirement for the new infrastructure that we implemented two years ago were essentially high availability because the data we are backing up to disk needs to be available even if we use the data center because this is the main memory of the organization. It needs to be scalable, we have to cope with the data increase that we are facing on a daily basis. We need to make sure we are able to manage the cost effectively because we cannot afford to pay twice every year in terms of licensing, in terms of storage, in terms of tapes, and in terms of manpower. We need to keep in mind that this storage infrastructure needs to be managed by people. So we had a different perspective when we implemented this new backup solution. And what we did is, in fact, purchased two appliances. We put these two appliances in two different data centers and we have been able to achieve the complete backup of the infrastructure within our backup window without any problem because now we are able to backup the entire infrastructure in approximately 12 hours. Then the next 12 hours we replicate the data to the second data center and the next 24 hours are used to duplicate the information from disk to tape. So in this 48 hours backup window, we have the full cycle and not only we are backing up 70 terabytes of data, but we are duplicating twice the information. So we are managing more than 200 terabytes of data within the weekend with the infrastructure that we put in place. So you said if you had to do it over again, you might have done it differently. You made some mistakes, but that also sounds like it was largely the organizational requirements changed. How much of it was, if I had to do it over again, I would do it differently versus changing requirements that you couldn't predict at the time? Can you talk about that? In fact, when we moved from the first infrastructure to the second one five years ago, our objective was really focusing on technology, saying we have to go to disks. We don't want to deal with tapes anymore. It's not cost effective. So focusing on technology was our main mistake because we have to also see this implementation from a completely different perspective, making sure we are able to manage the cost, manage the data growth, and manage the people. So that's why we did it differently two years ago. It was not that much of an organizational problem. It was more the focus we put traditionally on technology where we need to shift a little bit the focus on the people and on the policies. We are an organization where we produce a lot of policies that we were lacking in in terms of data policy. So Patrick, growing up in the household, with the head of household as a heavy IT practitioner, when I talk to your father a couple of times, people process technology. Technology is the easy part. People in process are always the hard part. Absolutely. Is this a common theme that you hear? So I can remember having to go down to the data center at the travelers when I was about 10 years old because the banks of the heart of the Connecticut River were going to flood and actually having to remove tapes from the data center that hold basically the general ledger for the information systems because of a basic disaster scenario. So it's real and it impacts people. So some of the technology we come out with new technology all the time and try to innovate, but really at the end of the day it's about SLAs, especially for backup. You want to make that window and those windows are getting more aggressive every day. We have people who are classifying more applications as mission critical. So given the tools to not only help people assess what their SLAs are and what they should be, then providing technology to back that up. And then for backup it's a lot about cost. So we try to help customers with ROI studies for store wants and some other interesting financial vehicles, like capacity on demand to be able to put some more predictability into the infrastructure. And the big cost is people. So if we can get this down to very efficient from a cost perspective and an operational efficiency matrix, it's really helpful for the customers because backup can be difficult, but you don't want someone sitting there and dealing with backups 24 by seven, right? You want it more of a function of the infrastructure. So Frederica, how much virtualization you have going on in your organization? Virtualization? Yeah, your system's highly virtualized? Yeah, we started the virtualization journey in 275, which was not progressing very quickly. It was initially, essentially, about test and development systems. And we changed completely in 2009, and we went aggressively on virtualization. Now we have approximately 83% of virtual environment. So how did that affect your backup? Yes, of course, it had a very high impact on backup because it changed completely the way we were backing up all these environments. I would say it changed the way we are doing backup a little bit initially, but we are still continuing doing backup in the traditional way by putting agents inside the machine and backing up the virtual machine as it was a virtual machine. And we are moving more and more to a kind of snapshot technology where we can check the VM as it is without going into the VM and use the new functionality that are provided by the different players to be able to back up much more quickly the entire environment. And we are very excited about what has been announced yesterday. So how do you do it today? Is it a weekly full and a daily incremental? In terms of backup, we are trying to stick with these weekly incremental and weekly full as we were doing in the past, but more and more we are moving towards a full of the time because it will be easier in the future to deal with potential restore operation, especially if we have the main desktop. And you'll be using snapshots as a means of doing that. What kind of RPO and RTO? In fact, right now we have critical applications. We have identified through a business impact analysis, 10 critical applications where the recovery point objective is about a few minutes and the recovery time objective is about four hours maximum for these machines. And for these particular applications we are not relying on backup more on replication techniques to make sure the information is always available. For other applications less critical, I would say we heavily rely on backup to be able to restore the information, but the RTO is more around two days, five days. Okay. And then tapes, you use tapes as a last resort offsite? It's really the last resort. We moved tapes outside to different locations in case we have to go back to tapes. It happens from time to time because with this environment we have a retention period about two to three months. But sometimes people are coming back to us saying, I had a file three years ago, I need this file because I need to base my new study on the information that I got stored in this document. So we have to go back to tapes from time to time. And you replicate the 6200 date on the 6200 as a disaster recovery, but it's synchronous, right? A synchronous location? No, it's asynchronous. It is, okay. We backup and then we replicate. Okay. So you're quite a distance, the two data centers quite far apart. And then so the tape is really a very last resort compliance. I hope I never go there. One of the benefits that we've been able to achieve with this last implementation, get away from tapes to be able to focus essentially on this and with this technology. Now, how about HP? Obviously, you're an HP shop. So you've got an affinity for HP, but how has the experience been specific to the backup project? From HP, from the simple HP support, the product itself? We had a very, very good experience with HP specifically on the backup side. HP was very interested in our case study, in our scenario. They helped us a lot in implementing, finding first of all the right architecture and implementing this relation to the backup software that we are using, because we are not using HP Data Protector. And we have some kind of very special partnership with HP, which allows to have a communication channel directly to the HP apps. So from an HP and communication perspective, it was a great opportunity. You're not using Data Protector yourself, what's your backup software? We are using another backup software from another bandit. Okay, you don't want to say? Yeah, a backup from Simon Tech. Yeah, okay. A lot of people use Simon Tech. This is the world we live in, right? Anything you want to see HP work on? What's on their to-do list from your standpoint? Of course, as we are not using an HP software in conjunction with the hardware, we like seeing HP working on the HP catalyst to provide a good interaction with the backup software. That's what they are doing. And they have been announcing yesterday a very cool feature more in the Simon Tech Terms accelerator, which is transforming a 12,000 backup operation into a few minutes operation. It's very cool. We've been able to test it. It's very cool. So we need to continue to invest a lot in interacting with software from other vendors. It's really critical. Well, you've been with HP a while. So we always say with the cube, we've been documenting the turnaround of HP. When Meg came in, she said it's going to be a five-year turnaround. We're in year three. From your standpoint, from a customer perspective, how do you feel about where HP is as a company at this point? I think HP is going in the right way. They are still focusing a lot on what people have right now, their current problems and issues like, once again, what have been announced yesterday by David Scott and the keynote, providing more and more functionality to existing hardware or to a platform that we know and we like managing. But at the same time, they have this different perspective in trying to define this new style of IT by changing a little bit the way companies are operating their IT shop. For us, we are not re-talking anymore about IT. The IT department is not called anymore the IT department. We are in the digital transformation journey. And I think HP will be a great partner for this journey because we need to focus more and more on the business and less and less on the IT itself. So we are here to provide a service not to introduce technology. Hi Patrick, how about you? How do you feel these days? Yeah, some of the comments really resounded with me. You know me, I love me some technology. I love good products and innovation. But if you look around in the conference today, you see a different taxonomy here. You see things like new style, transformation, you know, business focusing on business outcomes. So it's definitely, I think that HP has got a really good DNA in innovation, especially on the product side with HP labs and things that have been grown here, things that have come up through acquisition. We've got great technology, but focusing on sort of transforming our customer's business, infrastructure is hard, right? You can't expect someone to go rip and replace all of their products all in one fell swoop. And, you know, we don't just sell widgets here, right? So it's about the journey, it's about the transformation, it's about providing the people, the services, the, you know, the help along the way from an implementation standpoint. So for me, it's been a big transition of very sort of product focused and, you know, innovation on the product side to, you know, really moving towards better business outcomes for our customers. So that's a good bumper sticker, but I'll give you the last word Patrick. So the truck's pulling away from HP, this is my partner John Furrier loves this question, but what's it say? The bumper sticker from your perspective, you know, your business, what's the message here? Another 75 years, right? I hope I'm here to see it. Yeah, you will, you will. As long as you don't go to any more storage parties. Excellent. Well, Frederick and Patrick, thanks very much for coming to theCUBE. Yeah, thank you very much. All right, keep right there, buddy. We're going to the keynotes next. Should be very interesting. I think I'm guessing Martin Fink is today. I don't know if any of you guys know that. Yeah, so, oh, David Scott next. Oh, sorry, we're not ready for the keynotes. Well, that's right. He's coming a little later. That's how I saw people find my bed. David Scott's coming up next. Always, always exciting, always interesting. We'll get his perspectives and then we'll go to the keynotes. Keep right there, buddy. This is theCUBE. We're right back after this word.