 Welcome, we're back, this is Dave Vellante of wikibond.org, and this is theCUBE, SiliconANGLE's production of OCP Summit 5. theCUBE goes out to the events, extracts the signal from the noise, and we're here with Wesley Jess, who's the VP of Supply Chain, along with his colleague, Aaron Sullivan, who's the director and principal engineer of Supply Chain at Rackspace. Gentlemen, welcome to theCUBE. Thank you very much. Good to see you guys. It's not often that we get to interview folks from the supply chain side, but it's really relevant for this conversation. Wesley, let's start with you. Talk about your role at Rackspace. Yeah, I think we've got a very unique design in our supply chain. Our supply chain is very untraditional. It has a lot of your traditional functions, but in many cases, untraditional functions, all the way from license management, all the way to the hardware engineering infrastructure team actually sits in our supply chain. So Aaron Sullivan, principal engineer, is integrated within our supply chain. So if you think about integrating sourcing, the relationships we have with our supply base, the tier two component players that we interface with, all the way to our integration partners, to what we license and free software, and actually operating systems as well, sit in supply chain. And having a tightly integrated group like that actually helps us move very, very quickly and come up with very innovative designs at a very fast pace. So your supply chain kind of mirrors your stack. Is that a fair way to look at it? Very much so. The only thing that's not inclusive in it is really the networking side of the business. And my counterpart, who oversees data center, also oversees the networking side. So Aaron, how did you end up in supply chain? What was the path that you took there? Did you come from the product side or? I did. I worked in the product end of the business. At one point, a few years ago, I had some ideas for ways that I thought we could move faster, we could be more effective. And Wes was already at Rackspace at that point. And I encountered him, had been working in his group for a little while, and we had a discussion about things we could do to move better and faster. And a lot of it was already in that recipe book that he has that he just described. So let's talk a little bit about the OCP journey. Maybe take us back to the beginning. And how it started from terms of your involvement and what it's meant to you guys and where we're at now. Okay, great. Take first crack, Aaron, and then I'll pass it over to you. I think it starts with the OpenStack, frankly, our engagement with the NASA and OpenStack and basically kicking off that open initiative. And it was a very easy parlay for us to move that right into the hardware side. The most interesting part in the early part of our journey, we've been doing it probably two and F years, is the amount of hardware players who told us we could not do this. Now there's nothing like telling a racker that you can't get something done. We got exceedingly focused and it's not like, I think some of the themes have been, it's not our motto of people. We could literally fit our entire hardware engineering staff inside of a van. It's a pretty small group, but Aaron was engaged very early on and does a lot of the community work. But we found really leveraging off of what was OpenStack based and moving that to OpenCompute was a very natural migration for us. Anything you wanna add to that or? Yeah, I think if you, I mean, you see it here at the show, but if you look at how OpenCompute has changed over the course of the last few years, I think our engagement with it really helped to add some influence to the model where we said, look, we're really excited about what this brings. We're really excited about the potential of this community, but we also need to expand it to meet the needs of companies that are more like us, the customers that are more like the ones that we have. And in that process, it started with just a few people we continue to grow the engagement and the influence to say, here's how we think infrastructure needs to expand to encompass this ever larger community. And you see that some of that showing up in the designs that we showcase today that you see out on the show floor that we worked with some of our partners. Well, so we're running a crowd chat, a crowd chat.net slash ocpsummitv for five. A lot of people don't realize this is the fifth gathering, it's dramatically increased in size, even just from last year. So it feels like, Wes, we're sort of on this S-curve, you know, there's an old guy for all you mathematicians out there. So where are we? I mean, is this just about to explode? Is this the sort of the year that OCP really breaks out and becomes mainstream or do we need a little bit more time to bake? I think so, that's been our experience. We had worked on a number of designs that we planned on deploying in mid-13 and actually we actually brought those production level designs working with two, it's actually, we went in Quanta, partners in the OCP community and we're ready to deploy them on scale and then, well, behold, our customer base came back and told us something about our product offering, needed to be different. So we had to very quickly shift the offer that we had in the infrastructure and Aaron and the balance of our engineering team within a very short order, roughly three months, we were able to refactor the designs that we had and utilize probably 70, 80% of that design and within a three month period and now we have, and that was in the fall and now we actually have on scale a production ready scale volumes landing in Q1 of this quarter and then in Q2 we have a brand new product design coming out also based on a lot of the principles we learned. So we see it really picking up, not just with us but with a lot of the other players out there. So talk about how it's affected your business so far and how you expect it to change going forward. Yeah, Aaron. When we first got started, I think we mostly looked at Open Compute as something that we could use to save money, that's something that I think a lot of IT infrastructure people see. We saw the larger potential but the real drive here is we can not only do this but we can do it more cost effectively and I think, initially, that's the big drive for almost everybody as I think about summits over the last few years in the news. The headline was always I save this much money or I ran this much more efficiently. But we're now beginning to see extra capabilities that we can add and new technologies that don't really exist in the more closed proprietary systems, starting to come on board. Companies are actually starting to invest their money, their resources into this in ways where they might have previously done it in a very low exposure, kind of small system, maybe one or two partners. It's starting to get quite a bit bigger and we see that playing out a little bit more inside Rackspace too. We're starting to get partners come to us and saying, can you build me something that does this? I can't find it anywhere else. And of course, with this community and these capabilities coming, we can. So this morning, I don't know if you guys heard the morning keynotes, Frank got up this morning and he made the statement about proprietary BS and then on the panel that I was hosting I said that their converged infrastructure, all that proprietary stuff is a $400 billion market space. Now, people took that two ways. On one hand they said, wow, this thing's really going to get disrupted. On the other hand, they said, well, there's a lot of investment going in there, there's a lot of function. And I'm interested in what you just said about a lot of the capability that you're seeing in OCP-like architectures, actually extending beyond some of the proprietary designs. Can you give some examples? Yeah, I can. Let's start with just a couple of simple pieces. There are components that exist out there that many of these big suppliers acquire today and they sort of deliver it to you a certain way. They factored, this is how you should use this thing. And this thing really has a lot more capability than the way you're allowed to use it. But a call has been made, kind of like, we use this analogy a little later today, it's kind of like going out to buy a car. We designed it for you to drive it this way. This is how you should drive it. And we designed it with these options. So these are the ones you get. And if you come in and you say, yeah, but I could do more if you give me this a little bit differently. If you just change a little bit about what's there, I have so much more potential. It's really hard to go talk to a car company and get them to change a model, shipping year, and even some relatively minor ways to meet your needs. It's a great analogy for what we see as well. Some of what is in the server when you tear it down and you look at it on a bill of materials basis, it's not so much different than what's already out there. But maybe it's a custom firmware. Maybe it's a chip that's running at a slightly different speed. And maybe it's a way that those components are integrated into the larger system. But consumers of this stuff that get into those parts really start to see ways that they can tweak it and add value. And there's a huge value adder for us just in that particular aspect. Wes talked earlier about refactoring these systems. 70%, refactoring these systems. That other 30% is really going in and saying, okay, I need to change these pieces to deliver this new and different outcome. I don't have to reinvent everything. Just give me the parts and the freedom to change the things that I need. So it's really the flexibility aspects here that bring that into play. So talk a little bit more, Wes, about the customer feedback. So you had mentioned earlier, well, customers said we need something different. This sort of leads from the flexibility discussion. What were they asking for? How were you able to respond? And sort of what did you learn through all that? Sure. The rack space is all about fanatical support at the end of the day. It's how do we help our customers? It is how do we help them become successful? Which you can imagine, you get a lot of feedback, both positive and then ways we can improve. And with that, we run net promoter surveys. So we get a lot of active feedback around how well each of our product offerings are doing, positive or negative. And around this particular platform, we had a lot of feedback around performance. Really performance matters. The input-output performance is a big deal. And in essence, we had our cloud offering was a one-size-fits-all. And we determined, hey, we actually need to change that around. And so very short order, we actually introduced, we call them internally, different flavors. But they're different options for our customer base. So they could tune their application more toward that platform. And that was a very quick piece of feedback that we got. And we, as a company, I think pretty quickly shifted our strategy and put that product out there. We've seen some very positive adoption around it. So go back to the early days of OCP. Back then, the conventional wisdom in the media was that the big-scale internet players like yourselves, like Amazon, like Google, Microsoft, and others, were all about sort of one-size-fits-all with layer software on top of it. We've really done a 180 from that. Haven't we, from a design standpoint? I wonder if you could, Aaron, talk about that a little bit. What changed and why? Yeah, it's a pretty big world. And there are a lot of big players that need something just like what you described. But you go back to that $400 billion enterprise market. It came up thinking and adapting to a certain type of software, a certain use case. And that pile of spend is looking to go somewhere with it. But they can't necessarily pick up and just jump over to this new model. They have to evolve. And I think we recognize that in our own customer base. So many different types of users that we have to address where we look at them and we say, we got to have something that suits each one of your needs. And all of you want infrastructure, you want to think about your infrastructure kind of like Facebook thinks about theirs. You want it built for you. You want it customized to your particular needs. Now of course we don't build a unique platform for every one of those. But we see this space in the middle between these very sort of traditional ways of delivering compute and these new, very forward thinking ways of doing compute. And we say, we'll be out here, but we have to bring these other users, many of which are our customers along with us. And we have to adapt to doing it the way that they need to do it as they evolve. So that's got to put tremendous pressure on the supply chain. Maybe talk about Rackspace's philosophy there and how you differentiate with the supply chain. Yeah, it's, talk about complexity. It adds a lot of complexity. So we could have, our public cloud was a lot of interesting learning for us. Because public cloud, we could look at utilization on Tuesday of one week. And then we could look at it two weeks later. And it could be dramatically different. Workloads could have moved around. So talk about the difficulty of, how do you plan for equipment for that? The one thing we learned is, you can't plan for it. So what you got to plan for is to have a very elastic supply chain, a very elastic partner base. You can respond to big ups, wings. And traditionally, that's been a challenge for the company. But through our relationships and our partnerships and part of the organizational structure I talked to you about a little bit earlier, we've been able to, I think, pretty effectively be able to respond to the internal needs. Of course, they always want it faster. But if I look at the prior traditional models on how we could execute versus how we execute today, many, many fold faster. And that's in a big part due to our relationships with our supply base. We had Colin Crawford on before, and he made the statement essentially that rack and stack servers are a dead end. They're kind of a road to nowhere. And he talked about, you got to start thinking about racking racks and even maybe racking data centers. I wonder if you could comment on that. You guys, I presume, share that philosophy, but where are we in terms of being able to? We do. I think that's happening. We've definitely seen that happening in our business for quite some time. There's a change in the way that your customers think about doing business that influences this. And I think, you know, a few years ago we started seeing these sort of solution in a box, in a rack type of ideas coming out from all kinds of providers. And it's exciting right at first, but you got to think about how you're going to fit into that, how you're going to use it. And as soon as you identify the value of it, and our customers kind of do this in segments, as soon as they identify the value, they start to ask for just that kind of thing. And of course on our end, just like Facebook was saying, just like Cole was saying, it's an efficiency thing. The more we can deliver in that particular aspect, the easier and the more simplified that piece of our supply chain is. So you guys have talked a little bit about, we've made references to this server a couple of times. Maybe talk about what you mean by this server. And share with our audience a little bit of, unpack that a bit. Sure, sure, so our open compute designs are based on slightly modified windmill boards. We place a greater emphasis on manageability and serviceability in the data center, especially online serviceability than what you see out of the typical designs. Now, when I say we, what I really mean is, us in collaboration with some of our design partners, they do most of the engineering work. You could think of us as a very engineering savvy end user that can speak their language and understands the language of our customers and can communicate what it is that we need out of those systems. So we place more emphasis on management, systems management that more traditional enterprise shops are familiar with, and that many other users don't really mind if it's there or not. It might not always add all the value, but it's sure nice to have it there when you need it. And then we place a lot of emphasis on a really, really powerful IO subsystem in the rack. I think if you look at our rack designs, if you walk out on the floor and look at some of them that we collaborated with, I think there is more networking bandwidth, more drive bandwidth packed into that rack than any other open compute design that I've seen out there right now. So talk about the requirement for that IO subsystem to be advanced as data grows. You got to move it around, but talk about that in a little bit more detail. Sure, sure, sure. Well, I'll do a little pitch for one of our latest products. There's 40 gigs of networking to every customer. Every customer that runs on a machine has 40 gigabits and it's highly available. So switches can go down, networks can go down, components in the server can go down. It stays online. They not only wanted all that performance, but they wanted it so that the odds of it going down are very low. They wanted the same thing in drive bandwidth. Of course you have all these solid state disks. One of the things you should start to realize is, man, how do I take advantage of all that performance? Some of these disks are so much faster than the storage systems they're connected to that unless you can find some ways to uncap that value in those systems, you end up stranding some capacity. So what's historically the data center has been just a huge consumer of labor. I mean, it still is. But it sounds like you're attacking that labor problem with engineering and supply chain excellence. So what happens when something breaks in your environment? Do you just let it go? Something fails over? Do you eventually just take the rack and stick it into a wood chipper? Why don't you describe that philosophy a little bit? Sure. We treat it all the way down to the server. Server goes down. In many cases, that means a customer is down. So that customer, they could have HA or high availability, failover, some don't. We have so many different types of customers. So we treat every server that if a server goes down, it's customer impacting event. So we do everything that's possible within our data centers. Depending on what you talk about, what you throw in the wood chipper or not, that depends. For example, philosophically, we don't let drives leave our data centers. Anything that would have customer storage or customer data does not leave our data centers. We melt them. We actually, strangely enough, and I oversee our corporate social responsibility and sustainability as a part of supply chain, but we actually center punch them with a actual, just think of punching it, and then we actually crush them. So the opportunity for any sort of data recovery is very, very low off of those drives. But we govern if we go across all the different components that are within our data center. Philosophically, it's about keeping our systems up and going. And one of the things that Aaron touched on, he talked about our server and one of the biggest differences is the mechanical designs, the amount of time we spend on the mechanical aspect of the server design, to be able to feel what we call hot swapping fans. If a fan dies, some companies don't care. We care. We want to be able to actually keep the machine running, swap the fan out, put a new fan in. So think of if a drive dies, if a fan dies, if some component dies, you want the balance of the server to continue running. So we incorporate a lot of those design principles. So a server doesn't go down. And that's really kind of core to our philosophy. Awesome. All right, so I'm going to leave you guys. I'm going to ask you one last question and I'll let you think about it. I'll give you a hit, bumper sticker. Okay, so think about the bumper sticker as where tomorrow we're pulling away from OCP summit, summit five, pulling out of the San Jose Convention Center. What's the bumper sticker on this year's OCP summit? You want to take a break? Get on board because it's happening. Get on board, it's happening. Yeah, get engaged and we'll address that in our presentation later this afternoon and we will help people understand how. So let's tease that a little bit. Your presentation this afternoon is, what's on? It's at two o'clock, two o'clock this afternoon and it will both be Aaron and I and we'll talk a little bit about the journey we've been on and even touch on some of the points we've chatted about. Awesome. All right, Wesley and Aaron, thanks very much for coming to theCUBE. It was great having you and good luck with the initiative and congratulations on all your success so far. Thank you very much. Thank you, Dave. Keep it right there, everybody. We'll be right back. The Silicon Angle, we're live from the San Jose Convention Center at the OCP Summit Five. We'll be right back.