HPE ProLiant Compute Gen12 Portfolio Overview

So, my name's Darren Anthony, and, uh, I'm with, uh, the product management team. I lead the product management team for the rack and tower servers. Next few minutes, we'll just, uh, spend some time talking about, uh, the ProLiant portfolio, uh, but a good, very excellent segue.

We're gonna tee it up and talk a little bit about history. We, we like to reminisce and talk a little bit about the past. And so let's just talk about ProLiant.

So, 1993 was when the ProLiant family was actually, uh, announced. Uh, and it's been a tremendous journey of innovation and exciting products, and a lot of, uh, if I can say it this way, blood, sweat and tears have been invested, uh, across the company, uh, whether it's from our sellers, uh, from our solution architects, from our engineers, uh, to product management. And it's just been an exciting journey.

So let's talk about it just a little bit. So, 1993 was, uh, the debut of, uh, the first ProLiant, uh, rack, uh, rack mount servers. Um, I, I'm gonna give, give myself away a little bit, uh, of, of, you know, my tenure.

Um, I started at, uh, the company with the c in front of it. I started, uh, with System Pro, which was a predecessor that was, uh, one of the first, uh, multiprocessor servers that was on the market. And so that's where, uh, where I got started.

And, and, you know, I've jumped in just full, full on because I love technology and I love the innovations that have come from HPE, from Compact and HPE. Moving on, uh, we've got in, uh, 2003, uh, which was, uh, the first four processor, uh, four processor ProLiant server blades, which was also exciting, very revolutionary. Does anybody remember when blades when, uh, blades began to, to ship and the excitement?

So, I'm gonna dig way back. Does anybody remember, uh, P class? I see a few people.

How about eClass? Ah, yeah. So if you remember eClass, then, then, uh, you, you've got a history here, so very exciting.

And, and then we actually turned a corner in 2012, which was where, uh, proline Gen eight came on the scene. Lot of excitement, lot of press, and a lot of focus around that as well. That was super exciting timeframe.

Uh, moving over down the timeline, we have 2017 was actually where we were called out to be the first vendor to build Silicon Root of Trust. Anybody remember that timeframe? Yes.

Very, very exciting. Does anybody recall that? Everybody wanted to copy that and then start to use the same Silicon Root of Trust messaging that we were using.

Uh, you know, that became the, the, the, yeah, us too in the market. Moving over, uh, we, uh, launched in 2022. Another exciting, uh, um, uh, family.

Another set, uh, set another exciting set of, uh, platforms in the portfolio. This was Gen 11. And then most recently, uh, has anyone here not heard of Gen 12?

Does anybody know that? We announced Gen 12? Okay.

Yes. Very exciting. And yet, again, building upon that 30 plus years of innovation and advanced technology, we've done it yet one more time.

And once again, now, I would be remiss if I did not go to the next slide. Very busy slide, but very exciting slide, because this shows the progression of one of our key and very proud intellectual property capabilities inside Prolia. And that is our IO and we have just continued to build upon it, continue to innovate.

I not everybody realizes this, and I, I like to highlight this, and hopefully everyone here does. But that is an HPE asic that is our asic not bought off the shelf, but it's something that we control. We design, we wrap the software ecosystem around it, we wrap the software experience around it.

And we are the only OEM in the industry that can say we design our own ASIC for manageability and for security. And we've continued to build upon that and continue to innovate. And then we fast forward to where we are right now with Gen 12, and now we have taken it to another level with the capabilities that we have put inside, uh, of the ILO and the asic.

And it's a super exciting time to be at HPE to be selling HPE products, to be talking about HPE products and talking about, uh, HPE servers. Let's talk a little bit about where we are today. So, gen 12, I won't go too deep into this because we've been through, uh, a lot of this already with all the messaging, the announcements and everything.

But it is very exciting about where we are today and how Gen 12 is based on the three key pillars. So we have that next level of security, which is really what's driven by, uh, I, our ILO and a lot of advancements that we put in that for Gen 12. We also know that in the world that we're in today, that applications are continuing to evolve and continue to demand even more and more resources.

And so we had to continue down that path of what we've done previously and continue to optimize on our portfolio and continue to deliver as the app new evolving applications come on the scene such as ai, which are demanding more and more resources than ever before. And we are delivering with ProLiant Gen 12. And then finally, no secret here, that, uh, data centers and IT managers, they need to do more with less.

They need more automation, they need more capabilities. Now, it is not exempt from being incorporated and being required to provide information as it relates to, uh, contributing to corporate standards or corporate, uh, uh, goals around, uh, power reduction around, um, sustainability, reducing carbon footprint. And so we've also incorporated a lot of that in our software experience as it relates to the ProLiant portfolio as well with our compute s management, our comm offering.

So in that next level of security, this is a great place to talk. I showed the slide on the IO and kind of the progression of IO. So we're delivering on a multi-layer of security.

And the one thing that I am super excited here about the IO is that yes, we, we built the I, uh, the ASIC from the ground up, and that's our asic. But in gen 12 with ILO seven, we've added this secure enclave, which is delivering yet another level of security, uh, for our customers to protect our customers. Uh, I, I like to borrow a phrase from Scott Schafer because he referred to the, the Secure Enclave, which is a separate security processor in the, in the asic.

He refers to this as like, ILO has its own TPM. Now, I'm gonna use a word that I, I like to use, and I, it's a lockbox. It is a vault for a place for storing and separating from the system of s storing passwords, cryptographic information as it relates to that platform certificate information in order to deliver that next level of security.

And then it is no surprise that the threat actors are nonstop, they're busier than ever. They're continuing to look for ways to exploit, continue to, for, for ways to get into a data center. And one of the ways that they continue to do that is how can they, how can they get into an environment through the hardware itself?

And we know that quantum computing, as it becomes more available to, to people more like us, rather than being reserved for those really, really large, uh, research facilities, as quantum computing becomes more prevalent, more available, that it's actually, it's being used for good, but it's also being used for bad. The threat actors are always looking for another way to exploit. And so we saw that, and we made the effort in order to, to harden our system and to be ready for, to be quantum resistant.

As those threats begin to take on, as those threats become more real, that we're ready for that and we're ready to, to, to keep quantum, um, quantum attacks at at bay. And then trusted supply chain, this is another added level of, of comfort that we bring to our customers. Because whether it's in the manufacturer where the parts are being built, coming to our factory, to the factory floor, being loaded on the truck, and then being delivered out to the customer site, customer wants that comfort level of knowing that every step along the way that they've been protected, and that they know what is in the rack or what's in the server and what's not, and what shouldn't be there if something has been added along the way.

And then they can track it and keep tabs on it and give that extra layer layer of, uh, security and protection for those customers. So, super exciting, uh, about what we've done with Gen 12 and about the long history we've had with the, the 30 plus years. So here's a little more, more about the IO if you haven't seen the slide.

Great, great slide talks about, uh, that, uh, ASIC and the secure enclave. And, um, just really tremendous. The one thing I want to point out here is, uh, prior to IO seven, uh, that was a 32 bit asic, and again, being, wanting to be ready for the future and look for what we do next, is that we actually move towards a 64 bit asic, uh, for IO for ILO seven, which gives us a whole new level of playing field of capabilities that we can bring not only when we launch Gen 12, but the capabilities that we can bring going forward and be ready for those new threats and the new things that are happening within the market from a security perspective.

And then as far as the, the, uh, tamper protection, I've talked about that a little bit, uh, already, and this is where that, that separate security processor, the secure enclave comes in really adding an extra layer of security to protect our customers and add additional security level within their environment. So that's a, that's a single, single diet, I assume. Yeah, it's a single asic Yes.

But You have within the asic a somehow walled off The processor that you Yes. On. Yes.

But you know, if, if you, we can talk a little bit more about it if you wanna pause here. 'cause there's actually somebody in this room that's from that team that knows the security of IOI Think we're getting to dive a little bit later. Okay.

You are great. I Just wanted to Okay. Clarify that, that point.

Yep. So you're focusing a lot on security in this Discussion. Um, how important is security to HP's customers?

Um, you know, I think a lot of people think that servers are very low level, that the software on top does all the security and so on. How much, um, how, how, how much of a requirement, how high on the, on the, the level of requirements from customers is security, especially supply chain security and, uh, the, the, the hardware that they're running on. It's a top of mind.

It, it is. There's not a conversation that goes by that none of us have within the product teams where customers are not concerned about security. And what year, year after year, the security, it mo it bumps up to a higher level of concern for customers just because of the, the threat environment that we have.

Mm-hmm. So, uh, you know, I'm not trying to say they're not important, not, Right? No, no, no.

But, but, but specifically hardware security Yeah. And things that are being addressed by the ProLiant line, like Right, like the secure enclave. And you also did mention supply chain.

I want to, I want to hear right. You know, how does HPE guarantee the customers that the, that these servers are, that the integrity of the server is, is Good? So first, let's, let's take an analogy.

So you build a house, if your foundation is cracked, then how solid is your house gonna be? So that's a good, that's a great analogy. In the case of compute, you're only as good your security story and your, your security for your environment is only as strong as the underlying foundation that you have.

And it starts at that. We all know that compute is an underlying foundation for everything that happens in the data center. So you're, from a foundational perspective, you're only as good as the foundation that you've established.

And so that's why it is so critical, just from a hardware perspective. 'cause the last thing you want is for there to be some entry point that's at the hardware level. Now we also know that the threat is all across the environment.

There's software, you know, there's networking, there's a, there's a whole bunch of other areas, but where the data path resides is the reason why it is so critical, and the data path is actually coming from the server itself. So, right before we go, because that is a pretty interesting topic to set up, uh, which is handling this at scale. So as we're thinking about this as a, from a governance perspective, system administrators may not care anything about whether or not a chassis was opened or as, you know, are constantly changing out parts, et cetera.

This is the, you know, the secure enclave is alerting me to this. How are you engaging like the CIO versus the, the VP of infrastructure when it comes to security and the ecosystem as you're about to go into? How is that conversation different, if at all?

Well, I don't think it's a different conversation. I mean, I think it's just, it's related to the overall, what the security concerns and what the security strategy is for the environment. Um, but I'm not sure if I'm answering your question exactly.

Yeah. So in a really large organization, the VP of infrastructure cares about security. Right?

But the CISO is the one who makes sure that he cares about he or he Right. Cares about security. Yeah.

So it is a ecosystem. The CISO might operate the knock and the VP of infrastructure doesn't, right? So the, while the VP of infrastructure may not care that a chassis was opened or that a part was changed out, other than making sure his system stays up, the system on the knot does care.

Mm-hmm. So kind of this, the, the tooling for the knock versus the tooling for the, and if we're gonna go into it deeper in the, in the later conversation, great. But I wanted to, from a ecosystem perspective, how you're approaching the needs of the ciso, who's specifically in his silo or her silo cares about security and security operations versus that of a VP of infrastructure, who's more or less checking a box.

Right? You're gonna go into that one deeper. Yes, sir.

All right. So you wanna hold that one For Yeah, I'll hold it for the, for that conversation. Okay.

All right. So a little bit more, this is a great slide that kind of, uh, matches what the, what the ask is there around, um, supply chain. So we know that at any place within supply chain, there can be something inserted that can be a security threat to a customer's environment.

And so at HPE, where we enable this secure supply chain is be able to keep tabs at any place within the flow of sec, of a supply chain to be able to keep tabs on what's happening with that hardware or potentially that software. In the case of, you know, ilo, this is why we, we manage and control our own asic This is why we've implemented, uh, the secure supply chain, is to give customers that added peace of mind. Because the last thing you want to have happen as you're actually having servers built is where someone has maybe diverted a server off to the side, they've installed a chip clip, or they've swapped out some hardware.

Maybe they've made a, a change in some way, some alteration to the server. And so this is an opportunity from a secure supply chain perspective to alleviate that and add that level of protection. We've even enabled it where from the time that the truck leaves the factory, that the customer can track the truck and keep tabs to make sure that the truck hasn't taken a side road, and then somebody's, you know, opened the back like you see on some of the covert operation TV shows, or they've opened the back, swapped out some hardware, and then threw something else in inside the truck and send it on its way.

So it was just another layer, layer of security and comfort level, uh, that are being delivered to our customers. And actually customers love it to the points, which is the reason why we expanded it to the rack level to give customers that peace of mind if they're having a whole rack built, an entire rack built by HPE, that they've got that peace of mind. So does that answer the, the question?

Yeah. Actually, this is exactly my question and I really appreciate this. I No worries, worries, because This is exactly where I was going.

Yep, Yep. So now we move on to talk a little bit about, uh, performance, uh, and so, and optimization. So we know that more and more that the applications, uh, you know, many, many of our customers are not just doing a year over year over year lift and shift and keeping the same applications.

They're doing the same thing over and over. Customers need faster, better applications. They need applications that allows them to do more, be able to extract more information, to be able to deliver more value for their business, to be more competitive in the market.

And so, as applications are evolving, especially a applications like ai, as the demands are there as it relates to the performance of our servers, uh, performance of any server, that the optimization and the ability to be able to deliver the efficiency, uh, for those applications is what we're focused on here. So not only do customers need to consolidate and save, we'll talk about that a little bit, uh, but then it's really about being optimized for those AI platforms. You know, more GPUs, more capabilities, more, more resources, as those power hungry applications are being deployed in order for customers to move faster, do more, and to deliver more to their own businesses.

And then finally, it is inevitable that, and the path that we're on as processors, as GPUs are demanding more resources, they're, they're requiring more or, or are demanding more thermal thermals to be addressed. That we have to look at evolving the servers to be able to support liquid cooling or, or maybe to say it this way, alternative cooling rather than air cooling. We are reaching a threshold in a point where air cooling is becoming more of a challenge.

I have talked to many, many customers, and we've seen this in our own internal analysis, uh, as far as our, our units and, and the configurations that are going out the door and the configurations week by week, quarter by quarter becoming richer, customer had one customer said, be prepared. I will fill every square inch of your server with something, whether it's Dems, whether it's PCI cards, whether it's, you know, et cetera. And so we're seeing these rich configurations more and more that customers are building that then demands to be able to support alternatives to, uh, air cooling, maybe air cooling can't tank, can't cover it.

And then what's really great about this, and we can talk about it a little bit more in a minute, is we're able to tap into the rich heritage around liquid cooling and all of the, the, uh, patents that we hold at HPE, other C company, right? No. Yes.

So yes, yes. That was a C company. Yes.

That was an acquisition C company. Yeah. So around our ai, uh, optimized platforms, this is really exciting, uh, is that, uh, you know, we're delivering these, uh, jointly developed platforms, uh, with Nvidia.

And so, um, with gen 12, we're able to deliver two x, uh, more GPU acceleration, uh, compared to previous generations. We're also positioned to be future proofed and ready for what comes next, uh, from Nvidia as they are rolling out and launching higher wise GPUs, higher td, higher tdp, GPUs, GPUs that are demanding liquid cooling. We're well positioned for that.

And we're ready. So this is great that you, um, are working so closely with Nvidia and that you're future proof for them, but you guys have a pretty rich history with HPC and mm-hmm. ML and dl.

So, um, do you also have any AI optimized platforms that don't use GPUs? Because a lot of the really hard work that goes into some ai, um, workloads doesn't need a GPU u Yeah, it's a very valid point. So, uh, you, you know, what we saw with, uh, um, with Gen 11 was, uh, Intel really pushed, uh, one of our, our partners Intel really, uh, came on strong talking about the accelerators that are built into the processors themselves.

So you're absolutely right. Uh, you, you know, when you look at workloads, um, and uh, I'm gonna step out here a little bit and say that just about all the new applications have some element, whether it's small and insignificant to large, just about all the applications are taking advantage of some AI capabilities. Everybody can't, you can't have a GPU in every single device in the world.

Uh, there's no GPU on my phone, but I'm able to take advantage of some, some co-pilot and, and, uh, make use of some other, uh, capabilities. Uh, or I'm even hearing about, uh, some folks running little small, um, LMS on, uh, um, on their phones. Wow.

No GPU on the phone. So it's a great, great example. And so, so back to Intel.

So yes, we are, we're, you know, that's one of our key partners. Obviously we've got Intel platforms, and so we, uh, you know, we take advantage just like as part of the architecture, those accelerators are built into the Intel processors. And so we're able, uh, to deliver on those capabilities that may not need a full blown dedicated discreet GPU, but yet still need, you know, some type of, some level of capability, uh, from an AI perspective as it relates to accelerators.

So, yes, and going forward, anytime that, uh, you know, any of our partners, whether it's A-A-A-M-D or whether it's Intel, if they're driving that into their architecture, we're gonna take advantage of any partner with them on that. Cool. Thank you.

Great question though. So it's a great observation as well, uh, because a lot of people don't realize, uh, that there, there are other things that you can do, you know, like with those built-in accelerators, but it is the key about the accelerators is all of the applications having the right hooks to be able to take advantage of those, of those accelerators. Great question.

Is There a, just follow up on that question. Are there architectural shifts you see, I mean, you, I see the Gen 12 has really more than evolved over 30 years. You've gotten smarter better.

But what about the challenges, the density and the compute and thermal wave? Is there any materials that you can do differently inside the build of the server that's cost effective at scale that would help you leave you that thermal envelope that you're challenged with? Because the design for these are modular to, to accommodate a lot of customers, but once you know that you have a certain thermal footprint for a certain workload, wouldn't there be some optimizations in the chassis or optimization and maximizing airflow before you go to that bigger step and that big switch to cooling?

Yeah, so that, that's an excellent question. I'd like to say it this way. We have, we started introducing some in Gen 11, some of our platforms with, um, liquid cooling capability.

Again, it's a choice point. It all depends on how rich the configuration is that customers or want they need to deploy. But we were able to get to this point with the fantastic engineering that we have and the innovation that engineering does, that they've been able to delay us this long with being able to support air cooling.

Well, what we're seeing is two things happening in the market is that not only actually there's three, uh, and I'll talk about the third one, but what we're seeing is not only those rich configurations that customers are needing, as I mentioned, customers said, I wanna fill every square inch with something. So you have customers that are maxing out on the memory, they're maxing out on the IO with PCI cards, they're maxing out on the N-N-V-M-E drives or whatever the choice point is. And so all of that just adds to the, the number of not only power consumption, but also the TDP.

Now you bring in processors that are, you know, pushing 330, 350 watts, you put two of those in there and you see it adds up very quickly. So it all depends on the configuration. That's, that's one.

And, and we're seeing those rich configurations, but also more customers are looking to, whether it's in the environment that, uh, they need to address some regulatory requirements. And so they need to make some changes to their ambient temperature in their environment. Or maybe, uh, the customer has target goals, uh, for sustainability for reducing their car carbon footprint.

So it's actually three things. So it's that meeting regulatory requirements, maybe customers need to do that. Maybe they're trying, just trying to reduce their power consumption, you know, for sustainability, or they're just requiring those really, really rich configurations in their environment.

That's what's driving this. It's not so much a point of that. We don't want to continue to do air cooled and we will, but where things are going and we're making choice points and continue to do that around, you know, selecting the right materials and, and optimizing the air flow.

Engineering has been doing that for the last 30 years and excelling at it. It's just that all these other elements are coming into play Now, is There, you think there's a limit we're at towards the end of the road for air cooling because the density demands? Or, or are we still just predicting that we'll be sooner than later?

I mean, Uh, I, and I think there's still gonna be room for, for air cooled. Um, but you know, when you're pushing things like 600 watt GPUs and you put eight of those in a server, it it becomes a, A jerky factory. Yes, exactly.

Becomes a challenge. So it, it's just, it's kind of the progression and the direction that things are going and, and where people are going and where the technology's going. Uh, but there will still be place for, you know, because not everyone can make the investment to have a full direct liquid cooling offering.

I dunno, there's a thermal balance between condensed versus, yeah. Yeah. Otherwise, so just curious.

But, but it is also where we're seeing more and more customers that are looking to, to move from a density A one U more into a two U because you have larger fans, they get more airflow. And so that's the other thing that we're seeing, because I mean, when you have a one U server and you max it out, you're starting to kind of go up against a brick wall. Yep.

Literally. Thank You. So, and I'm, I'm running outta time.

All right. Uh, so how do we wanna progress? I'm running outta time.

Got it. All right. Portfolio slide.

I love this slide because it kinda shows the breadth of the Gen 12 portfolio. Uh, so what is available now, actually everything's available now, that's, uh, from the middle off to the left. And so we have, uh, our two AI platforms, uh, our DL three 80, a Gen 12, as well as our DL 3 84, uh, gen 12, which is, um, uh, based off of, uh, the Grace Hopper, um, platform architecture.

Then in the middle we have, uh, these were announced in February. And, uh, we're actually, uh, we've been, uh, open and taking orders on those. And so, uh, this is our full portfolio, our DL three 20 and DL three 40.

Our one socket platforms, A one U2 U flavor. Uh, and then going down the list, we have more of the density optimize, which is our, uh, two socket platforms, which everyone knows the DL 360 and DL three 80 platforms, uh, which are the, the, depending on which month you're in, whether the number one or number two, selling platforms, uh, in the world. And so we're very excited about those platforms.

And then finally, uh, our, uh, smb uh, slash edge, because it's a tower platform. This is our ML three 50 Gen 12, uh, and the two socket flavor off to the right available, um, coming up, uh, here this summer, uh, which will be our Gen 12, uh, on the Synergy four 80. Uh, and then, um, our, our, um, database, uh, INM Memory Database, workhorse or Workhorse, which, which is our, for Big Data optimize, which is our, our, uh, DL five 80 NA, um, four u uh, four socket platform.

That's all right. Questions, and then cannot, uh, leave this out. But this shows the full breadth of the portfolio that we have, including our, this is our Gen 11 platforms.

And so we've got, uh, just a, a broad offering of platforms depending on the workload, cho workload choices that you have. And, uh, it's just really, really exciting to be a part of HPE.