In The Arena by TechArena - A Breakthrough Sustainable Cloud Infrastructure Collaboration with Ampere and HPE

Episode Date: June 13, 2023

TechArena host Allyson Klein chats with Ampere CPO Jeff Wittich and HPE VP Peter Groth about their recently announced collaboration for cloud infrastructure optimized for compute sustainability.

Welcome to the Tech Arena, featuring authentic discussions between tech's leading innovators and our host, Alison Klein. Now, let's step into the arena. Welcome to the Tech Arena. My name is Allison Klein, and today I'm delighted to be joined by Peter Groff, Vice President, CTO, and GM of HPE, and Jeff Wittich, Chief Product Officer for Ampere. Welcome to the program, guys. Why don't you go ahead and introduce yourselves? Peter, why don't we get started with you? Yeah, so hi.

Yeah, I'm Peter Broth. I work in the computing space inside of Hewlett Packard Enterprise. And, you know, my roles here are looking after the service provider market. Very specifically, I run that business alongside a bunch of really talented men and women. And yeah, I'm a general manager of that. And also I serve as the chief technology officer for Compute Worldwide. And Jeff, why don't you go ahead and introduce yourself? You're a return guest to the program.

I am. Yeah. Thanks for having me back, Allison. It's always fun. I'm Jeff Wittek. I'm the chief product Officer at Ampere. If you haven't heard of Ampere, Ampere is a modern semiconductor company. We're building the first cloud native processors for the sustainable cloud. And we'll talk a lot today, I think, about the innovations that we've been driving. It's all about high

performance and efficiency delivered via the cloud and really even expanding beyond that to all compute. Now, HPE and Ampere were on stage together at OCP Europe. You were talking about some joint innovation that we'll get to in a second. But before we get there, Jeff, I know you've been involved in OCP for a really long time. Can you just give us a sense of how this organization has changed since its inception? Yeah, like you said, I have been involved in OCP for a long time going back, I guess, about a decade or so. And there have been a lot of changes that have been really exciting to see, I think in a couple of areas. One, OCP has really taken on a broader reach over the last decade, wider mix of OEMs, more and

more hardware partners. Also, it's gone beyond just the big hyperscalers into other clouds, enterprises, even the edge is really getting into the mix as well. And then you can see with the event a couple of weeks ago, just that expansion beyond just the US. With the regional summit in Europe, you can see more and more people participating in the open compute community. The other thing is just from a technology perspective, that reach has also broadened. So it's not just system hardware

anymore. They're also looking at things like chip level interconnect and data center RAS. There's a lot of really interesting work streams. The other thing is the recent focus on sustainability over the last couple of years. That's also been a really interesting work stream and one that we've been very active in. Peter, HP has a long history of driving data center infrastructure innovation. I mean, your system's power, data centers all over the world. How do the OCP configurations help speed your innovation? And how does this work in OCP fit with your broader strategy for the market? Yeah, I mean, I'd like to chime in a little bit on what Jeff was saying, right? I mean, OCP has gone a remarkable transformation probably in the last decade, right? It used to be, you know, here is where all the, you know, snowflakes of individual hyperscalers basically

got published. It was almost, you know, just a spec review kind of place where you could jump off and go and bit that out to contract manufacturing. And it's really taken a change and a turn to something that from an HPE standpoint, we see as a much more practical and applicable kind of thing. You know, whether that is in the systems arena with things that we have been helping to drive things like DCMHS, which is the modular hardware system, but also in some of the areas where, you know, we have a very, we have a very vested interest. that would be things like management systems management, especially, right, we have been very heavily involved in the open

VMC environment in DCSM-CM, which is the modular stack for or modular hardware to be able to plug in. So from that perspective, we see OCP, you know, probably as the modern style of a standardization body, it's a little bit more democratized than what you would see in your typical IEEE or, you know, some of the more traditional kind of standardization bodies, simply because it's not just a question about having a bunch of manufacturers sitting and discussing what's best for them, but frankly, bringing in the consumer side of this as well and saying, well, what is actually really good for you? And ultimately, A, it helps with speed.

It helps with innovation that's, I would say it's more relevant. It's super relevant. And a bunch of the other stuff has been relevant over the years as well. But especially when we look at many of the OCP-inspired customers and customers who adopt OCP-inspired or designed kind of infrastructures, this brings to a point a bunch of customers at the same time with speed and with purpose to build what it is they're looking for and not necessarily just a bunch of things that's great for the vendors themselves. Yeah, that makes a lot of sense. Now, I mentioned earlier that Ampere and HPE discussed a strategic

collaboration at the show. Jeff, why don't you just start and talk about this collaboration and why it's important to those customers that Peter was just talking about? Yeah, Ampere and HP, we've been collaborating a lot over the last few years. There's been some platforms we've put together. There's a great ProLiant RL300 server out there right now that Ampere and HP have worked together on. Specifically at the OCP regional summit, we talked about the work that Ampere and HP are doing together around DCMHS, that modular hardware system that Peter was referring to. So we're both contributing to that effort. There are a number of different subgroups under that DCMHS umbrella within OCP. The advantage for customers is that they're going to be able to

get hardware quicker with configurations that better meet their needs, which is going to allow them to have a lot more choice than they had previously. So this is a really good example of Ampere, HP working together, bring some really cool innovative technologies out to the customers, ones that are actually helping them solve their real problems, which are speed, agility, as well as greater choice from a technology perspective. Peter, why don't you add your perspective on the collaboration and talk about why Ampere was a great partner for this, and how does that extend what you're offering in the market to customers?

Yeah, I mean, you know, first and foremost, right, I think that HPE's hallmark has been innovation, you know, frankly, in a sea of followers for the last 30 years, right? If you look at many of the things that we take for granted today, simple things like RAID, right, you know, that came out of our laps. You look at, you know, our very first management solutions, right? We actually call it the RILO. It's since evolved to the ILO. But, you know, nonetheless, you know, systems management is something we take for granted today, you know, yet I remember when I started my career, right?

We were sitting and doing this on, you know, serial cables with consoles and all sorts of wonderful things. Yeah, I know that that speaks, unfortunately, a little bit to my age here. But, you know, the reality is, there is, there comes these, these, these waves of innovation inside of the server technology environments. You can say the first one was, frankly, when we took what was arguably a PC CPU and turned it into a server. Our very first DeskPro was built around the 486. Yeah, that's a very long time ago. But there are these waves where you see a step function in innovation. And frankly, we felt like this was the right time to go out and look at the ARM-based environment. And Ampere was an obvious partner in that. They've gone through, they've done a lot of the maturation that needed to be done in the space, ecosystem, making sure the run times are

there, that they're optimized, you know, that it's not just a bin packaging kind of environment. And so from that perspective, you know, when we look and we purvey through the technologies available, they're really typically they fall into kind of three categories. There is the myth. It's kind of like, yeah, it's nice, but you know, does it actually solve a problem? Then there is the, oh, it's nice, but does it actually solve a problem? Then there is the, oh, it's solving a problem, and it's probably solving for 15% or 20% or 25% better than something else. And then from time to time, you see these big step functions. And frankly, we saw that with Ampere in terms of especially in and around sustainability performance per WOD, you know, and with the ecosystem being available around it and being in a very mature state, you know, we try to

and we try to arm based architectures about six, seven years ago, eight years ago with our moonshot product. You know, it's a great product. The problem was that the rest of the ecosystem around it just wasn't there. And it clearly wasn't there. And it clearly is there today. And that's a big testament to Ampere and a bunch of the other companies out there innovating in the space. And so for us, that made it very obvious that this is value that we had to bring to our customers. It makes a lot of sense. Jeff, you talked about sustainability earlier. And when I was in Prague, I was just struck at how much sustainability was a topic on people's mind. We've really seen a sea

change in the last couple of years in customer focus on sustainability, baking goals related to that into their bottom line. How has this shaped your strategy at Ampere for cloud-centric silicon? Yeah, I agree. We have seen a massive change in the market. And what's great for us is that from day one, efficiency and sustainability, they've always been core to our mission. So it's really great to see that the rest of the market is placing a lot of emphasis on that. And I think one of the big shifts is that a couple of years ago, when we talked about efficiency, it was generally just about saving cost. Or maybe with some more mature organizations, there were some goals around sustainability or ESG goals that they were trying to meet. But oftentimes, it was a bit of an afterthought. And now it's really become core

to what everybody's doing. Not only are they trying to be more responsible and save money, but as data centers consume more and more power, it becomes more and more difficult to continue to procure and provision more power. Building new data centers is really difficult to go and do and isn't really the responsible way to approach the problem. What we've seen is that its efficiency has become core to how clouds and other enterprises actually build out compute. At the end of the day, they're all limited by the amount of available power. And so it becomes critically important to them that they utilize that power envelope and deploy the most compute as possible. So it's not really just a sustainable cloud being about social and corporate responsibility or cost savings anymore.

It's about the inherent ability to even meet your future compute demand. And that's why we've focused on what we focused on at Ampere, which has been building both the highest performance computing, but also the most efficient computing as well. So it's no longer a choice. You can have both. And that's why our whole product design has really oriented around maximum performance at scale in an efficient manner, which is just a really different design point than what the industry is focused on over the last couple of decades. The industry is really focused on high performance, but always at the sake of efficiency and power consumption. Yeah, and I think that's actually, I want to pick up on that just a second here, Jeff.

I mean, I think that's a really key point, right? I mean, thus far, for many, the whole ESG sustainability goals have been around a series of public statement about, of course, we've got to have to look out for the environment, the planet, what have you not. But, you know, what I think has been the big watershed moment and where I am here has helped as well is, you know, frankly, driven by a macro

that's a little bit different, which is we're fundamentally as IT. You know, if you look at technology today, it is an inherent part of society, right? There's very, very few things we can do unless that we have technology. Our fridges don't work, right? You know, phones don't work, right? We're back to pen and paper.

The cars don't operate, right? You know, it's pervasive throughout society. And if we look at that statement and then at the same time also understand that we are consuming a very vast amount of the resources that we have to offer, right? I mean, I've seen anything from 8% to 14% of all the energy in the world in what is being generated out of the power plants basically being consumed by IT in some aspect. And when it comes to the data center, you could argue that there is at least 10% of the world's resources of fresh water going in there as well for the cooling. And fundamentally, what we have to understand as an industry, whether we are on the supplier side or whether we're on the demand side, is fundamentally that we are competing for these resources with other industries. And if you look at that from a macro lens,

that means that when we go into a municipality and say, well, we want to build a data center there, that municipalities have spent an enormous amount of taxpayer resources putting in whether it's roads, whether it's roads, whether it's electricity, whether it is telecommunication, power, infrastructure, et cetera, et cetera. And typically that would attract industry like, say, for example, a steel mill. That would generate localized economic activity, right? There's direct jobs to building it. There's direct jobs to building it, there's direct

jobs to running the steel mill, and there's secondary and tertiary industry around it that also generates economic activity, which ultimately results in tax revenues to be able to pay for all this infrastructure investment being done. Data centers are fundamentally different, right? You build a data center and sure, there's a bunch of construction workers helping setting it up. But after there, there's like, well, 10 people, maybe 12 looking, you know, after the facility. And it doesn't actually generate

any localized tax revenues. So we have to be able to change our mindset into one of how do we do a lot more with a lot less? And I think Ampere has really helped in that space. You know, it's not just Ampere that's on this, but I do think that Ampere gets very high marks, both from us, but also certainly from our customers, that, you know, they're helping tremendously in that challenge.

How do you stay within the resource envelopes that you already have and that you frankly fought hard to get? Because try to get yourself a hookup license for power or for water anywhere in the world right now. That's a hard, hard, hard problem to solve. Now, Jeff, I wanted you to take a little bit of time to respond to Peter's lovely words about Ampere to give some broader context of what you've done within your portfolio, you know, beyond what you've said so far, because I think Peter's brought some additional topics up. And then, Peter, I'm going to follow up with you on what HPE is also bringing to its very robust portfolio in this space to help customers with full sustainability. So, Jeff, why don't you go first? Yeah, what we've really done, like I mentioned, with our core mission statement being delivering high performance at incredible efficiency,

what we've really done is we've completely broken from the legacy paradigms. So legacy x86 CPUs have focused really on maximum performance and the cost is power. We've taken a completely different approach. So we've created a complete line of processors, our cloud native processors. Ampere Ultra processors are the ones that are in market today. And they have a completely different design point in the sense that we designed them so that they are both high performance and efficient, which meant that at the architectural level, at the complete processor design level, the way that we do pre-Silicon design, the workloads we look at, the way that we trade

off performance and efficiency in the entire design process is radically different from the way that legacy CPUs have been designed. So we've fundamentally brought a new type of product to market. Now, the product runs all of the normal software that somebody runs in the cloud. It's easy for someone to adopt it. But our approach is totally different. And the result has been that when you look at real cloud workloads, we can deliver something like two and a half times more performance at the rack level than x86 CPUs. And that's rack level with, you know, same power envelope, same space envelope, same water consumption, right? Everything's the same. We're just delivering multiple times more output compute. So when you look at some of the

problems that Peter was mentioning, the fact that building new data centers is incredibly hard, if not impossible today, what do you want to do then? You want to make sure that every rack in your data center is fully utilized and that it's delivering the maximum amount of compute. So that's sort of the radically different approach we took was maximize performance at scale and you don't get more power to solve that problem. Peter, do you just want to add a bit on how HPE is adding technology capability to the table to help customers with this massive challenge? Yeah, I mean, I'd love to. You know, I think there's multiple different things we're doing, right? I mean, first and foremost, we got to sometimes

take a step back, right? You know, the story in and around Ampere and the opportunity that it brings is obvious, right? I would go as far as to say that. I mean, you sit down, you read the literature, it's fairly outsized. The thing though is that there is a transition that's going to have to happen. And for as much as I as a technologist get excited about Ampere, Ampere is not for everyone. It's for many, but it's not for all. And that's because, frankly, when we walk in and we look at real world IT problems, say we walk into a financial institution, many of them are still running mainframes and mini computers and what have you not, because they have legacy code that's sitting on that.

And it doesn't transfer that well. However, the world is moving through this whole digital transformation, retiring a lot of that technical debt, which frankly we've built up over the last two, three, four decades. And as that happens, here has an opportunity and presents an opportunity for these customers

to break free of the mold. And so the way that we from an HPE technology-wise go in and build on that to lead to better sustainability, to lead to better choices is frankly, you know, by offering just that choice, right? And, you know, whether you're sitting with an x86 based system or whether you're sitting with an Ampere based system, make sure that the environmentals around it, right, the way that the product behaves, that that is consistent so that adoption is not limited or is impeded by that. So a great example of that would be, you know, I mentioned it before, our management technology, our services portfolio, the reach that HPE has, which irregardless of which silicon, you know, stripe that you nail your colors to, is consistent and the experience for the customers is consistent. system. Things that we started this conversation about OCP, things like bringing OpenBMC out into

the market, making that a first-class citizen across, whether it's x86, whether it is Ampere, across our entire portfolio, and literally going and saying, well, we recognize you're going to have a heterogeneous environment, but that doesn't mean that you need to have a heterogeneous experience. And so, you know, going and fully embracing the openness and frankly, the innovation that the community brings, that's super powerful for us. And one of the reasons why A, we partner with Ampere as closely as we do, but also why we view forests like OCP as an absolute key measure of how innovation is going to be moved forward. I know that we're coming to the end of the interview. I want to ask you both one question about the future and what you're seeing in the market more broadly. As we head into the second half of 2023, can you give a flavor of what you're most excited about in terms of innovation targets and what you

think is important for the industry to make progress on together as we continue to address customer requirements? Jeff, why don't you go first? Yeah, the two big areas. One is the one that we've talked a lot about. It's really continuing that focus on sustainable compute. Especially again, delivering more performance while consuming less power. And making sure the focus is on that

versus entirely focusing on just mitigating the results of runaway power consumption from legacy technologies. The mitigation piece, it helps, but we can't focus entirely on mitigation. We actually have to reduce the amount of power consumed per unit of compute. That's really the only way forward. So we'll continue to drive that across the industry with our cloud data processors, the current Ampere Ultra family processors. Our roadmap continues to move along at a really rapid pace. So you'll see some really exciting other products from us throughout the course of the

year. So we'll continue to drive that paradigm shift where you can have more and more compute without consuming more and more power. The second area of focus for us is really using the public cloud as that incubator for technology innovation. Peter mentioned who the early adopters are. And it's true, when you look at people that are running big public clouds, people that are running open source software, people running the cloud workloads, those are the use cases that we're really focused on today. Those are the technology leaders. That's where we'll be able to really focus on where the puck's going versus looking backwards at a bunch of legacy type of deployments. But what's great about that is that rapidly the whole world is moving to that type of deployment model.

So what today might be sitting in the public cloud is rapidly moving into enterprise private cloud. It's moving out to the edge. It's moving out to all these other areas. So while we talk a lot about public cloud and what we're doing in cloud workloads, you'll see us at Ampere really broaden that focus across more and more compute. As more and more people are taking advantage of those same leading technologies that sit in the cloud today, you'll start to see that spread everywhere. And our partnership with

HP is really a big piece of that, given how much of the overall compute market HP addresses. Peter, how about you? What are you seeing in the second half of the year in areas of innovation and collaboration? Yeah, I mean, you know, two things as well, right? The first thing is that, you know, and I touched on this earlier, that the sustainability story is now also an economic story.

And they haven't not always been aligned right the fact that we can clearly see that taking and utilizing and adopting technology that provides significantly better performance per watt performance at the rack level Etc like for example Empire is doing you know that that doing, that has a financial result as well. I mean, you obviously can't pay your shareholders with good intentions. They sort of kind of like expect a return on their investment. And the same thing goes for whether it's enterprises or cloud providers. Adopting new technologies allow them to

basically be more competitive or return more to their shareholders. And I see that as a really interesting space of development, which is really going to mature here in the second half, because there is now the ecosystem, the technology, and the ability to go and adopt these. And there are going to be winners and losers chosen based on the speed of adoption. And it's very clear. So that one is one that I'm excited about. You know, the other one that I'm excited about is frankly, the fact that, you know, having been in this industry for a very long time, you know, compute is kind of like a compute is the new pink, so to speak. It's coming back into core innovation.

It's not just bricks that you put in your data center. A lot of this is driven, whether it's from the big public cloud players or, frankly, from the enterprises who, in some cases, are beginning to repatriate parts of their stuff back. And as they're doing that, they've been used to costing levels in the cloud that they now have to go and replicate back home. And that's not easy, right? It requires that you go out and you graft a bull by the horns. You look at technologies like Ampere. You evaluate what your staff is doing. And, you know, just seeing the joy that are in many of the customers' eyes when they get back into the race of innovation and being, you know, not necessarily on the bleeding

edge, but certainly on the leading edge. It gets a lot of people excited and it's fun. It's fun to be in the industry right now. I wouldn't say it's a renaissance of compute because I don't think it ever declined, but it's come back at the fore of, wow, we have some very strategic choices we need to make and the technology decisions we make will directly impact whether we can deliver on those strategies or not. That's great. Thank you guys for being here. Really insightful interview. One final question for both of you. Can you please just share where folks can engage with your team and learn more about the solutions you're delivering in market? Because I'm sure we've piqued some folks' interest as they've listened online. Go ahead, Jeff. Yeah, the best place to check out for more information on Ampere is just go to

amperecomputing.com. You'll see a range of information on there, everything from product information about our processors. You'll find a lot of guides for how to get started, how to optimize specific pieces of software. You'll also see a lot of solutions there. So you can actually see some real life examples of things running. Lastly, there's also our regression suite is up on there as well. So if you want to check out over 150 applications that run every single night in our own test environments, as well as across a number of public clouds, you can go there, whatever software you're likely to want to get started on, it's probably up and running and you can go and check it out easily on AmpereComputing.com.

Sweet. And Peter, go ahead. Yeah. I mean, so I'm going to sound like a parrot to Jeff. I mean, obviously, you know, start out at HPE.com. It has, you know, the full complement of HPE products and services, as well as our award-winning GreenLake offerings for managed environments. But, you know, I also invite you to connect with any of our 68,000 plus partners across the world. We have a tremendous reach. There are great men and women out there that can help you through discovering new technologies and how HP can help. And if you want to speak directly to me or to my team, I invite you to hit me up on LinkedIn or to go out through our contact pages on hp.com. I know it sounds, well, when I press a

contact link, you know, does anyone actually read that? I can guarantee you we read every single email that comes in. And if the phone rings, we obviously pick it up. So, you know, by all means, whether it's with me directly, with my team, with the company at large, by all means, please engage. Peter and Jeff, it was a real pleasure. A ton of fun. Thanks so much for being on the program today. Thanks, Alison. Thanks, Alison. Thanks for having me. Thanks for joining the Tech Arena. Subscribe and engage at our website, thetecharena.net. All content is copyright by The Tech Arena.