 Hi everybody, we're back, this is Dave Vellante with Stu Miniman, we're with Wikibon.org and this is theCUBE, SiliconANGLE's production of this year's VMworld 2013, we're live at the Moscone Center in San Francisco. We're in Moscone South at the street level, come into the lobby, take a right, you can't miss us. VMware has set up this awesome space for us, really can't thank them enough and all of our sponsors here this week. This is the software-defined storage deep dive, it's a spotlight that we're doing on software-defined storage, SDS, Wikibon, Stu we call it software-led storage, software-led infrastructure. Chris Greer is here, he's the chief engineer of enterprise architecture at FedEx Services, which comprises IT, marketing and sales, is that right Chris? Welcome to theCUBE. That's right, thanks, glad to be here. Yeah, pleasure having you, really appreciate you spending the time here. This is a great event for practitioners like yourself, 10 years, is your first VMworld? This is my fifth VMworld. Fifth VMworld, so it's a show that was started for practitioners and it still has that first time on theCUBE. You feel to it, so I'm sure there's a, well how do you spend your time here? I'm sure people say sharing ideas, stories, talking to peers, I imagine similar for you, right? Yeah, absolutely, networking, meeting other people in similar situations, kind of taking their take on how they've solved problems, seeing all the different vendors and seeing what's coming to market, seeing what's new and seeing how this will play and how it might benefit one of our many different kind of areas that have problems that we have in IT. So let's start by talking about FedEx, I mean it's an amazing company and you guys are just going to continue to innovate. We hear the post office is getting pressure to stop delivery on Saturdays and you guys in Amazon are trying to get the same day delivery and Sunday delivery and it's just, it's phenomenal how that logistics complexity that you guys manage and how it's affected everyday lives with e-commerce in particular but talk about some of the drivers in your business, some of the challenges that you face, some of the pressures that your executives and your lines of business are putting on you. So as far as challenges go, Amazon has been in the e-commerce explosion has really been a huge benefit to our business but contrary to what they offer there's no such thing as free shipping. Yeah. So we're still working on cost as always a concern and as we grow out and as we are an international organization, we're global, we serve as 220 countries and territories around the world and the challenges of technology and that kind of a space and all the different regulations and things we have to deal with, it's a very complex working with an airline and a huge trucking company, a retail presence and we have all of that. So those are the big challenges that we look at. How do we do architectures that can be similar and bring this together so we can have one seamless experience for our customers. So what do you try to be best at? You got cost, you got availability and you got speed and agility. Can you be the best at all three of those or do you have to make tradeoffs? It's a difficult challenge but that's why we have the organization that we have where we look at these and we try and balance them. Of course we need speed, that's kind of baked in our DNA but reliability, we can't sacrifice that either because that's what our customers expect of us. So we treat our IT, our culture is more around making sure that we can deliver that same FedEx experience in IT and deliver those services the same way we do our customers because that's what they depend on us. So talk about the services that you're involved in. Is it organizational wide? Are you sort of narrow on a particular line of business? Talk about that a little bit. My role is in enterprise architecture. So we sit back and we get to go and look at these standards and build the architectures for anything from hardware to software stacks and things that we look at using all throughout the business. My particular role has been lead architect for virtualization at FedEx since we've done this about six years ago. And going in that the problem scopes are pretty much where we use technology. Chris, I'm wondering if you cover virtualization for such a large environment what's your take on cloud and software defined data center as VMware has put out recently? Sounds like you're heavily virtualized, you've got a lot of locations. You probably act almost like a service provider to your internal organizations. Walk us through how your thoughts are as to those environments. For us, we've been in this journey for a little while. We are about 80% virtualized now at this point. And it's an interesting transition as we see this going through. Virtualization has really changed the data center overall. And cloud is that layer of automation doing that on top. And it's really kind of enabled our end users and lets IT kind of get out of the way of the business. It's kind of what we've positioned it as. We have got a big private cloud internally and looked at other options of what the other options are out there. You can imagine a business like ours, we are a bit seasonal. So there are options to maybe look at in the future, bursting capacity, things like that. And the notion of the hybrid cloud, we're not there. But those are things that we kind of consider and looking at directions that we could go in. So tell us a little bit more about your environment to the extent that you can. Like a billion servers and 80,000 people in IT. Share with us the, if you can paint a picture in whatever detail you can share. So as far as this organization, sorry, the infrastructure and servers, I know we've just crossed around 22,000 virtual machines. Considering we didn't have the first one in production three years ago, that's been a big ramp up. So it's been an interesting challenge there. So we've got about seven, eight thousand people in IT defending and we're located predominantly in the U.S. but we do have offices around the world. But for IT we're focusing a couple years in Europe and Asia as well. So it's a global problem and a challenge to coordinate all that together. And how, you know, what degree of virtualized are you? So we're about 80% there on this journey and we're tackling the higher, harder end workloads now trying to go after and keep pushing the envelope to see what we can virtualize. So are you pushing towards that 100% fully virtualized environment or are there certain pieces that you don't think will get there in your environment? I mean, I'm not sure we'll get there as an environment as large and as complex as we have but we are definitely trying to virtualize as much as we can. There are huge benefits we've seen in terms of agility, flexibility, savings, and power savings. Even how we run our data center in terms of the power implications of things like that has been a huge win for us. Can you talk us a little bit about what is virtualization done to your storage and networking pieces of the stack? Like everyone else, I think we've bought more storage due to virtualization which has been interesting. We saw this coming but it's something to keep up with and a growth trend. More storage, the networking, I think we're currently facing the same challenges a lot of people have. Virtualization puts more pressure on our network to move faster. Once the server has been virtualized and we can do a lot of our automation there, the same level is expected from networking, from storage. How do we do those same things and she's the same agility and speed that we have server virtualization with networking and storage. So it's been an interesting problem. So what applications aren't you virtualizing? You talked about before. Some of our larger, we have a lot of custom written applications. Some of our larger ones are not. They're proprietary to us and unique to our business and very critical. And some of our larger workloads, large database workloads, we don't feel like they're fit right now but they're something we're still looking at. So you want to run those on bare metal? They are today still. Okay, so let's talk a little bit more about storage and software-defined storage. We're trying to unpack that concept. First of all, software-defined storage to a practitioner, it's like you're laughing. It's kind of an industry buzzword but there is some meaning to it. Software-led, we would call it. So take us through that part of your infrastructure, where you're actually using, so you're a long-time left-hand customer. Is that correct? We've been using left-hand actually about three years. Not that long. Not that long, but that's a left-hand array. Yes? Yes, we've started with the physical implementations and the VSA has always been interesting to us. So we've used it internally. We started using it as a training mechanism. Said, okay, to bring a new storage array or a new storage platform in, you can't just have multiple storage arrays sitting around for people to learn on. But the VSA allowed us to go spin up virtual environments for the SAN administrators to go practice on and give them a sandbox to play in so that the first time they were doing these configurations of changes it wasn't like they were working on production. So it kind of gives them the net to practice with for their tightrope routine kind of thing. Okay. So that's where we started with that and we've looked at extending that in a couple of small remote locations. We have areas where traditional SAN is not very practical. We have a lot of remote locations given as many different areas as we serve in the world. And we looked at it for places like that and the interesting dynamic that we get specifically with left-hand is the same operational model. So if we have this in our data center and it's one way to operate it and we have it in a remote location the way you manage it and the way your operations are the same it's been a huge win there. Okay. So using it for remote locations and these are locations that are IT light in terms of skill sets. Is that right? We don't have a lot of skill set there in some places it can be almost a big wiring closet and things like that. So there are not a lot of staff there. Not a raised floor. No raised floor. Water cooled. Yeah. Weaver AC. So this is giving a chance just to go and put the VSA in a small location and say okay we can solve your problem and we can still keep you consistent in that enterprise vision of the things that we need some of the advanced storage services. You may need snapshots. We have the same replication so we don't have a lot of different technology even though the demands for the business are I need it to be low cost here I can maybe focus that low cost VSA or remote locations but you're not sacrificing availability you're not sacrificing and learning a new different operating model so we have people that can operate in different areas of the business by keeping the technology consistent that way. So when you go to that software-based storage in that remote office what are the requirements of the we call it IT light from a skill set standpoint but what are the skill sets that you need there? So most of those they are very light but in some of them maybe remote locations with file print services you know light workloads office type workloads it hasn't been anything very demanding at this point so as far as that we may have a couple smurvers in a wiring closet you know three servers sitting back there and we could deploy the VSA they are virtualized because they were looking to consolidate they had a couple physical servers and as refreshers came up for the server hardware consolidate into one platform still mostly like I said file print servers things like that but it allows them to run other workloads there if they need to and still get the same level of availability From a practitioner perspective shouldn't all storage eventually be software led I mean don't you want that to be the case or why have any hardware in there other than the server? I like that idea a lot one thing that kind of divorces the innovation cycle I think you know you can look at your storage services from the hardware cycle which is rapidly changing so you don't have to wait a three-year cycle or whatever for a hardware or a traditional storage array to incorporate some new technology that's out there the latest and greatest Intel core so your performance actually ends up being an interesting curve that we're seeing in terms of faster performance quicker just with the hardware acceleration that comes out due to Moore's law I'm sorry Chris, can you walk us through you know you talked to us that when you went highly virtualized that had kind of a growth effect on storage you know when you put the virtualized storage in place did that have any effect back on kind of your VM architecture and the rest of your infrastructure? Sure, one of the first big things that we see out of there is I think there's been a sawtooth graph I've seen a couple of times at the conference that shows you know when you're in kind of a traditional storage array you have a huge kind of upfront cost which gets lower over time but going with a virtualized storage approach and you're solving software defined storage that price curve ends up being a lot more linear every server you put in adds some network compute and storage so it ends up being a much more linear and price curve to allow it for the growth and when you're in a highly virtualized environment and you don't quite know the demands the business is going to place on you that's a great thing to have to be able to scale out and keep growing just by adding more servers and then also to have that in a controlled price point What about things like VM density and your power consumption you know how does that impact it? Power consumption is something that we look at very heavily especially in my area it's geared a lot more toward the infrastructure side so we've seen a pretty large reduction in power usage by going with a software defined storage approach part of that is you've already paid kind of the power penalty for the server and it's already drawing some power and you're not drawing a lot extra adding the drives is not a significant increase so that's been a very good win and we've seen basically we're already pretty high compression ratio so it supports the densities that we're looking at is the big thing Chris what's on the vendors to do list generally HP specifically kinds of things they can do to make your life easier things they can do to not tick you off so luckily we work very closely with them and have a great relationship them being HP or vendors in general well a lot of our big vendors we have you have a lot of vendors close relationships but looking at that I think the biggest thing and we see this emerging now is allowing for more automation especially the VMware did a great job with the servers and allowing automation but as we extend this improving the automation trying to help us with this rapid scaling and letting us work in this first-class world certain vendors I think are definitely on that track other vendors are kind of a little more hesitant to figure out how that disrupts maybe their current business model but really helping us along that area is probably the biggest challenge that we have right now or the biggest thing we work with our vendors on is how can we consume this in a much more automated fashion and have it deployed quicker Awesome Crystal listen thanks very much for coming on theCUBE and sharing the FedEx experiences and your practitioner expertise good luck going forward and here's to envisioning an all software defined world and Stu thank you very much for sitting in with me Alright everybody keep right there we're back, Stu and I will be back up to wrap John Furrier is actually on his way to AT&T Park we'll be there tonight broadcasting live so stay tuned to theCUBE on SiliconANGLE.tv we'll be right back after this