 Okay, we're back here in live in Las Vegas at the HP Gen 8 announcement. We're going to deep dive, and we're going to drill down on all the guests, all the topics, and I'm John Furrier, the founder of SiliconANGLE.com, and I'm here with my co-host. I'm Dave Vellante of wikibon.org, and we're here with Mike Kendall, who is a CUBE alum. Mike is a group manager of the intelligent infrastructure and options for the industry standard server group, which means that we're going deep on energy efficiency. All right, sounds good. Welcome back. Good to see you again. Hey, good to see you. Pumped up the energy towards the end of the day here. Absolutely. So we were given, so just to go around, and Mike, I talk about what happened earlier in the day, is that a lot of different conversations with folks from HP and analysts, independent folks, you guys got really good marks on the technology. And what really blew off everyone off, the charts was the energy side of it, the power, the cooling, and the space. Huge issue. HP's got the whole pod thing going on. We've seen that HP discover. This has been a known problem. You guys have a unique solution. So share us the, what's going on there? Well, you're right. So energy is a huge thing and space and how you cool all this, because nobody is doing less data. No one's doing less compute. And they're all trying to figure out how to put more compute, more storage, more networking into the same size data center, trying to fit within that power envelope and that space envelope and that cooling envelope. So it's something that we really felt we had to tackle. And so, and as I've told people, this is a game of inches type of thing. You got to go ahead and be efficient everywhere. You got to concentrate, can I run the memory faster but run it with lower power? Can I go ahead and get more efficiency out of power supplies? Can I figure out how to reclaim trapped power? That's basically, servers that are allocated 750 watts, but they're only really using 400. Can I free that power up to run other servers so they can fit in those racks? It's not as as racks as you go into a data center are not totally filled up with systems because they think they're at the edge of their power and their cooling envelope. So those are all the issues we work to address. So my favorite topic is operating systems. It was my degree in college, my undergraduate in computer science. And two, no, three years ago, I interviewed Chandra Conte Patel at HP Labs. And he was at the beginning and pulling all sorts of stuff together. A ton of work around sensors. And the topic we talked about was the data center operating system. And that was the notion that in order to have an operating system, you got to measure stuff. So one of the key things that you guys are doing is around measurement and data. Can you share with us some of the things that are happening around that? Yeah, so you have our concept of our 3D CS sensors that we've really expanded on. So when we did the G6 and the G7 servers, we went ahead and we put sensors down on the motherboard, approximately 30, 32 sensors on it so we could precisely measure the temperature so we knew how to scale the fans up or slow them down more precisely so you get good reliability but you also don't waste a lot of fan power. Now, what if we expand that idea to not only do that in the motherboard but do it three-dimensionally within the server? We get a lot more information that way. And then what if we go outside the rack so we're able to also go ahead and even understand the location of a server within a rack and be able to track that information. So we now have sensors that are actually in the rack that track what servers are in the rack, where those servers are, and that kind of information. So we've done a lot of instrumentation on that and then we've also made the PDU smart. So PDUs look like to a lot of folks is glorified power strips but if you actually put a brand in them- Just tell me what PDU means. Okay, power distribution unit. So to a lot of folks, that's where you plug things in but it turns out that that's another place where you want to track power very accurately what's going- Fluctuations, et cetera. And everything else like that but you also then want to know how you're hooked up because that's a raw reliability issue because you've got an A feed, a B feed, A size power supplies, B side power supplies. Sometimes what happens is you have people hook up their A and B sides only to one side. Circuit. And then suddenly, so it says we're going to work on A. They click the A side and all of a sudden they go, but we know we're hooked up to the B side too. They click it off, half the rack goes dark. Whoops. That's a bad deal. That's human error. Yes. That's one of the things you guys are talking about is this human error is a big piece. I tweeted earlier that's a big part of the announcement. Talk about some of that automation because this is really game changing. It's a unique to HP. It's a big part of this announcement. You know, you got the thermal discovery thing going on. That's pretty huge. So just drill down and give us more knowledge around how this is all working because this is a big tech feature that you have. Okay, so let's take like location discovery services. So the normal way you go ahead and track your assets is you go up to the server, it's got a tow tag on it. So one of the common ways is take a laptop like what you have over there and you have a spreadsheet and you go ahead and fill out that. This rack ID at this new location you type it in. Or if you're really sophisticated, there's a barcode on that tow tag, you pull out and you scan it with a barcode. Well, that takes a lot of time. That's a lot of effort. And then you've got to go ahead and make sure you don't make any mistakes on that as well because if there's a particular problem with the server, you're trying to identify it quickly. You don't know exactly where that's an issue. With our location discovery services, the sensors we have in the rack and also in the server, when you push that server and that server can immediately query the rack and say, what rack am I in? What you location am I at? Puts that into ILO, sends that off to Insight Control so that you're able to go ahead and see and it actually auto-populates a nice picture of the rack right there. And then from there, you can go ahead and correlate that to power and thermal properties, et cetera right there and correlate all that stuff. And so that takes a lot of time. The other thing real fast on the power discovery stuff too is normally when you're wiring up a rack and everything or you're making any changes, you got to go ahead and document that once again, either in a spreadsheet and a hand-drawn diagram, et cetera, we go ahead and auto-populate that information so that you don't spend any time doing that. So it prevents mistakes. It also saves a whole lot of time on doing all the documentation. It doesn't really do with that. It sounds too easy to me. That sounds easy. It just auto-populates, it's like auto, you know, boom. What's the set up and boom? Just plug it in? Because what we've done is we've put, you know, microprocessor technology of work. And your car's got dozens of microprocessors. We came up with the bright idea of putting microprocessors throughout the infrastructure and into the server. And so, literally. It's in our operating system again. Here we go. There you go. And it's distributed too. It's totally distributed. It's distributed computing. So, will customers be able to lower their power bill as a result of this, do you think, or is that just a pipe dream? No, they should definitely be able to lower their power bill. Because, so one of the things we did in the server itself, so you compare it to, from like a G6, let's say, to a Gen 8, we were able through things like smart memory, more efficient power supplies, 3DC sensors, just on the server alone, we were able to reduce the actual power usage. You know, take a G6 configured with certain amount of memory and hard drive, same configuration for Gen 8, and we can reduce that power by at least 10% right there. Now, you couple that with the performance increase, and you may see, at a minimum, 70% more compute per watt for, you know, in less space effectively. So you can definitely go ahead and lower that power bill. That doesn't even count the fact that in the rack itself, we've made the air flow more efficient. So you can lower the fan speed in the servers and also your computer room air handling system. Okay, now hold on, so 70% improvement in compute per watt, some large portion of that is just gonna come about because of Moore's law. So that's not gonna save me any on the power bill, but the balance will, right? Yes. Okay, all right. Although that will save you on the power bill. If you're looking at a given amount of compute and it takes, instead of this much, it takes this much to do it, you know, for the same power, well, then that's a reduction right there. But then on top of that. Yeah, I got more data, so I'm doing more, so it's off-setting. And we're happy that you buy more too. But what we've done then is we've made that more efficient. So in other words, you've got a- Your relative power bill will go down in relative to what you replace it. Correct, correct. I mean, obviously- If you're adding more power requirements. If you're gonna- It's ill-efficient. Right, right. But you look at the fact that we're able to do that about a 2X factor. So if you obviously- That's the key. If you can go 3X, then your absolute number's gonna go up. If you go just like 1.7 or 2X, then your power bill's gonna be flat. Right. So that's a pretty big, big increase there. And as I said, part of that is because we've actually been, it'll take a server and reduce its power by an additional 10%. All right, so I want to ask you a question. Wait, wait, wait, just one quick follow-up. So these innovations, if I understand it, are going to permeate into other HP infrastructure, into storage, into networking, is that right? Yeah, because it's the proactive insight architecture. So the fact that it's an architecture means that we have design criteria and we have specifications and documentation on how to do that. So the way you do a common-slot power supply and the way that you have it so that it's able to work with the intelligent power discovery, that stuff's all documented and is being picked up and used by other organizations in the company. Yeah, and then you get a lot of leverage on your supply chain. With that, right? Because you've got specs now, you can roll that out to the other groups. You get supply- From our standpoint, yes, we get supply chain leverage from our customer standpoint. Once you learn how to use stuff, it applies to a broader range of the infrastructure. Great, awesome. So that's good. Yeah, that's really good, I love it. Great announcement. So a lot of stuff here, okay? A lot of things, kudos from the analysts and from us. You did a lot of, HP's done a lot of little things in aggregate that really make it work well and that's essentially an operating system as we were talking about. What gets you most excited of this announcement? You know, what is the one thing that says, Mike, that you love, you go, wow, this is, to me, the game changer. I think it's the combination of the things we've done around making everything in infrastructure smarter. From putting a chip in the drive carriers, not only making the drive carriers smaller, so you put more drives in, because people need more storage, but making them smarter so that there's a big, like this says, do not remove when you have a rebuilding process going on, or you're setting up a RAID set, all the way to the location, discover, I think that's, you know, as people really get their mind wrapped around that, I think that's really gonna blow their minds that we have smarts in a rack, and they're like, isn't a rack just a rack? You know, that we actually have that, and you're able to go ahead and push that server and have them, the server know where it is, and then send that to insight control, and you can actually watch that picture, populate, you know, click that tab for physical location, and watch it, watch it populate, you should push them in. Great innovation from HP. Mike Kendall, thanks so much for coming on theCUBE. This is gonna be the last interview of our day. I want to say thank you for bringing a lot of energy to it. Okay, that was a great guess. Yeah, yeah, and I'm looking forward to the next time I'm here, so always happy to see you guys. A lot of love to the energy we needed you, and I'm gonna slot you in when we have lulls of energy, because you got bring the power in there. Okay, okay, do that, do that. HP smart architecture, amazing technology, Gen 8, thanks for coming on theCUBE. You bet, appreciate it, thanks.