HPE ProLiant Compute at the Edge

My name is Priya Sharma. I'm part of the Edge Compute Product Management Team. Uh, first of all, thank you everyone.

The Tech Field Day team, all delegates for coming here. And in today's session, we are going to talk about why Edge computing matters and, uh, how edge computing is different from a traditional computing environment. And then we are going to deep dive into my favorite deal, 1 45, our, uh, Edgeline server.

We just launched in September. And then, uh, two of my colleagues will join me today. Uh, Gary Wong, uh, product manager for Edge Compute Solutions.

He is gonna talk about how our HP solutions tackled with real world customer challenges. And then we'll have, uh, Chris Bradley, uh, director of technical se uh, enablement team who's gonna talk about, uh, uh, about our management software, secure, um, secured, uh, com, uh, secured software that can help with, uh, uh, uh, managing all the edge deploys, all the devices at the edge, uh, compute ops management. Uh, so, and then finally we'll wrap up with how everything fits into our, um, broader edge compute portfolio.

So let's get started. So let's talk about why Edge computing matters, right? So we'll start with the market first.

Um, IDC predicts, uh, the market for Edge Hardware and Edge Software solutions is going to be about 358 billion by 2027. That's a huge growth. And, uh, when we look at that kind of growth, uh, that's usually, uh, because of the driving deri driving demand for the latency sensitive applications, plus the need for data security and AI opportunities.

And then if you move on to this side, 77% of organization consider Edge workloads highly critical. What that tells me that these organizations, they're not just tinkering with Edge, they're actually running, um, real-time analytics at the edge and making decisions. So think about this way, I wanna interrupt for a second, Jack Poller.

Yeah. Um, what is HP's definition of the edge? HP is F Edge.

We consider Edge as a unique location. Okay. Which is typically outside of a data center.

Think of retail store, factory, floor, outdoors, any field. Mm-hmm. That's the definition of Edge.

So it's a unique location For you. Is it just basically anything that's outside of a data center is considered edge? Yes.

Okay. Yeah. So coming back to, um, all the data that kind of paint, uh, paints the picture of how edge computing is important.

Um, so these, uh, 77% of the organizations, they do consider Edge as, um, as critical workload. So, and they rely on the systems to run their critical applications. So if their systems go down, that will actually hit their bottom line and customer experience.

And then, this is an interesting one. 25% of the servers will be shipped out, uh, will be located outside of the edge, uh, outside, uh, located at the edge, sorry, by 2027. And that's an interesting one because when I think of about this one, um, I'm thinking one out of four servers will live outside of a data center.

And, uh, that's a massive shift in terms of how we think about the hardware, the management and the reliability. Right? Excuse me.

But, um, to Jack's point though, this, that number's like based on IDC, right? Yes. There's two kinds of edges according to those guys.

Correct? You've got the, you've got the far edge, far edge, the device, and then you've got Right, call it the metro or macro edge. Sure.

Which is, you know, outside of the core data center mm-hmm. It could be a data center, right? Right.

So, so are you, so will you, does your version of Edge com you know, comply with that? Or is, or are you really just talking about device, uh, on site, uh, at customer presence? So In my version of Edge, I'm talking about far sites, but you're right, there is a near edge, right?

There will be customers like, uh, content delivery network or, um, service providers who may run in their own secure facility, right? In a data center environment, very controlled environment. And that is still considered a s so you're right, so there will be near edge too.

So in this definition, I'm thinking about the far edge, thank you. Where it's in some, uh, remote customer location. And what we've seen, I think, in talking to a lot of companies that are building, uh, edge applications, is that, uh, the change here is actually the s word, the server word.

Because in many cases there are a lot of, uh, devices running applications at the edge, but they're not servers. They're actually not appropriate hardware, they're little desktops. They're sometimes, you know, uh, the intel nook kind of devices and so on, right?

And, and what I think people are finding is that in many cases, those, those devices aren't well suited to these harsh environments. And so we're seeing a whole new wave of companies developing really nice servers that are designed for that environment that offer a lot more hardware capabilities than these sort of, you know, repurposed desktops would. Right, Right.

Absolutely. Right. Yeah.

We, we do so see a lot of those devices, but then we also see customers asking for those bigger devices where they can do data aggregation and the processor data. So they do need something powerful there. So now that we have talked about the market, how each, uh, edge computing matters, now let's talk about the edge itself, like how the architecture looks like.

So in my initial years, uh, this is where I used to see where all the processing used to happen, the data center. And we think of data center as a very controlled environment. You see stable power, you see stable cooling, you always have onsite, uh, dedicated IT folks who are only steps away to fix things if anything goes wrong.

And then with edge computing, you see a distributed architecture where all the low latency apps, uh, they reside on premises. And when we think of on premises, uh, location, we are thinking about back of the retail store or a warehouse or a factory floor, basically anything that is outside of data center, and they don't have any IT support and they don't have the proper cooling or limited power. Um, so that's why we call that environment as an uncontrolled environment.

And then let's deep dive into what we typically hear from our customers when it comes to uncontrolled in our environment. So space is a big one. I pretty much hear this, uh, from every single customers.

Um, customers, like I was talking to a retail customer, and they were saying that they wanted to use an Edge server for POS analytics, and they were planning to put the server under their, uh, POS system, and it's a very, very short depth, uh, cabinet where they wanted to put it. So they wanted something small. Similarly, other customers, like a manufacturer I was talking to, they said that they don't, they want, they have no space for IT infrastructure, so they wanted to put their equipments, uh, you know, uh, or hang it probably on a wall.

And they had these big cabinets that was hanging from the wall, and they wanted to put everything over there. So, uh, bottom line, uh, a lot of customers have a lot of space limitations, and they're always looking for something compact that can fit into those type spaces. And then noise is another one that I hear a lot, mostly from, uh, our retail customers where, uh, they're planning to deploy their, um, compute solution in a very, uh, uh, human environment, which is, and they get a lot of foot traffic, and their employees are working next to it, uh, to the server.

So they always want something that's very quiet so that it doesn't disturb their employees. And then airborne particles is another one that I hear, uh, mostly from the manufacturing guys. Uh, these guys, they mention their environment as very, very dusty and, uh, it doesn't get cleaned up very often, so they're always looking for some sort of a filtration system in their sys in their server.

And then temperature challenges, uh, we hear a lot, uh, about, about these, uh, different temperature ranges from our customer. Uh, I was talking to this customer, uh, last month, and they were talking about video surveillance, uh, in their parking lot. And one of them said that, uh, sometimes in the summer, their temperature can go from 120 degree Fahrenheit to 130 degree Fahrenheit.

So they want a server that can work on those during those extreme conditions. And then security is another one. And when we talk to our customers about security, uh, they're not just talking about, uh, the data security, they're, they also want to hear about physical security of the system.

Um, so we are here, we are, we hear that a lot from our customers about how to make sure that, uh, some person is not gonna just take the server away and then manage management is pretty much, uh, that we hear from every single customers, because majority of these, uh, customer use cases that we see, uh, customers are talking about deploying hundreds and thousands of edge devices in different location, and they need some sort of a me mechanism to make sure they can remotely manage and update their devices from a, a single gui. So these are all the diverse needs that we typically see at the Edge. Okay.

So now let's talk about h Vion, DL 1 45 Gen 11. So a little backstory, uh, two years ago, uh, we were at Discover and, uh, Krista, our vp, she was talking to some of the sales folks and they reached out to her. They said that, um, they're hearing about, uh, different needs, uh, from their customers.

And, uh, there was a possibility that our current, uh, portfolio may need something different. And then Krista connected us, uh, with them. And then we, we sat down with them.

We talked to almost 30 plus customers, and that's how, uh, we kind of captured all of their listen, uh, requirements. And then we made sure, um, during the developments, uh, phase for DL 1 45 me, we showed that all of these requirements were gonna be met, uh, with DL 1 45. And then, so let's, let's start talking about, uh, the space constraints that I, that I just mentioned.

DL 1 45 is a very, very compact server. As you can see here, it's almost 50% smaller than your traditional data center servers. It's about 16 inches deep, and it's also full, it's not full rack width, it's about 15 inches wide.

So you can just slide it in into a very tight space. It will, it will, um, it, or you can just put it on a wall. It comes with a wall mounting kit also.

And then, So guy courier, um, so it's not a full rack width. I can see that. Yeah.

It can look like a half rack width. No, it is not. So, and that's what that little slug on the side is for?

Yeah, yeah. Is that, if you're gonna rack it, um, you Yes, it, it comes with rack enablement kit, like a brick, uh, yeah. So that you, you can also rack it, and then it also comes with a wall mountain kit.

So you can put it on a wall If you're not gonna rack it, but you wanna, you know, have a three node cluster. Sure. What do you to put on shelves?

What are you gonna do on The wall? So typically what I've seen, uh, customers with three node cluster, they have these tiny, uh, mi micro data center rack. Okay.

Yeah. And they just slide it in like two nodes maybe with a switch and a UPS. That's the typical configuration that I've Seen.

I'm, I guess actually my real question is, um, if that still conforms to a standard so that people can some, some sort of standard for, for smaller Sure. Rack. I'm not fully up on standard For that, but I think that It can, uh, this, the standard rack is about 19 inches wide.

Yeah. And this is 15. Yeah.

So it comes with a rack enablement kit. Once you install that kit, you can easily slide it into a 19 inch rack. I'm, I'm interested in the depth as well.

Um-huh. Is it 13 inches, please? It, this is 16 inches.

Oh, okay. Because I, I just set up one of those yesterday actually, and it's only 13 inches deep. So I guess I'll have to get a bigger one if I, uh, But if you, uh, if you take this dust filter that goes in the front, it will be around 20 inches.

Okay. The reason, I mean, our typical bezels, uh, they're usually like one inches, but we went with four inches because it's all front cable server, so the cables have to come out from the side. And we wanted to make sure that, uh, the bezel doesn't impact with the cable bending.

So that's why it is about four inches wide and they all come out from the side. We have fan now opening on both the sides. Yeah.

Well that, that's actually really incredible too, because again, I was actually literally setting one of these things up in a, a Edge data center yesterday, uhhuh, and, um, a lot of servers have cables front and back. Yeah. And, um, you know, you have to provide actually, extra right space, uh, on both ends.

Yeah. Having everything on one end, having cable guides, uhhuh, all this stuff really goes to the requirements, uh, that these environments have. Right.

And the fact that the thing is basically ready to, you could really just set it down without a, without a rack enclosure, is, is really valuable too. Yeah. Yeah.

Absolutely. And, and it be mounted, so it can be mounted vertical with your rack mount kit. You can, um, you can hang it in a vertical rack configuration.

Right. Okay. Um, but this particular bezel, it is optional because I've seen a lot of those use cases where they're putting it in a disa disaster recovery vehicle.

They have their own special kind of rack, and the doors come with a filter itself. So, so that's why we made this optional, because not every customer is gonna use this, but for a retail environment where you're putting one or one or two of these, they use this a lot. Yeah.

So what's the power consumption on this thing and 'cause I see the fans there look kind of small and Yes. Noisy, I imagine they're not. So power Consumption, max power is about 900 watt.

A fully loaded, uh, configuration is about 900 watt, But it's probably not anywhere near that under a normal configuration. Right. I mean, that's maximum, That's a maximum con configuration.

900 watt, and it comes with run in power supply. So you can have two, two of these or one of these, it will easily operate. And I, when I'm talk about fully loaded server, I'm talking about three N video L fours.

Oh, I'm not talking about that. Okay. No, I'm, I'm like without that, let's end even Without that, the minimum is about 3 50, 365.

I, I can go back and check, but I think I, that's what I've seen. Yeah. If it's not fully loaded, eight cores, maybe with one or two drives.

Yeah, it's about that. And, and you said a core, uh, which processor is it? Intel and a MD.

So this is a MD fourth generation, um, epic Sienna processors. Yeah. Uh, we are one of the only ones 4,004.

No, it is the 8,004. Oh, the 8,000, yes. Yes.

That's a good one. So yeah, so this is the 8,004 series, and they go from eight cores to 64 cores, and they're extremely powerful, yet energy efficient. And when I talk about energy efficient, uh, the coolest feature that I see in this box, you can actually run a 64 core processor, uh, at 131 degrees Fahrenheit, 55 degrees Celsius.

So that's how energy efficient they are. And they're also, um, very cost effective. So, uh, it's, it's kind of a combination between the 4,001, you said a four and the 9,004.

Yeah. So how do you handle the storage challenges of edge computing? I mean, we're not expecting, uh, I'm sure the electric group would love to have electric MP mini everywhere, but, you know, what are you guys, what are you guys thinking?

What's the, what's the group thinking for how, how do you meet the challenge of storage for the edge? Okay, so, um, the storage requirement, Karen, kind of varies by customer and kind of varies by workloads. Everyone has high avail.

Yes. Everybody is different. So what we did, so this particular box can support up to eight drives.

Okay. Two for boot. And right here we have the NS 2 0 4 2 4 boot, and then you can either do two, uh, small form factor drives, uh, which will give you up to 30 terabyte of storage, or you can also do six E-D-S-F-F drives the slim form factor drives, and you get up to 92 terabyte of storage.

Wow. That's a footprint. Yeah.

And, and this is something that every customer loves. I wish I can take this out. Nails.

It's great that you're using those new, uh, yes. 92 terabytes. See, so these are M twos that goes into that nine zero, the NS 2 0 4, they're, uh, about 480 gigs.

So you can get two of these for, uh, OS metering. So up to eight drives, Two terabytes. And then, uh, now that we are talking about server, so, um, it has four fans in the back with N one N plus one redundancy, which basically means if one of them goes down, the server will still operate without an issue.

And then, uh, it has PCI Gen five slots three, um, four, four PCI gen, I'm sorry, four PCI gen five slots, uh, three for GPR next, and then one OCP slot for controller or next. And then, uh, we also have, uh, the bezel filter I just mentioned, up to eight drives, two for boots, six for storage, redundant power supply units. We have the chassis intrusion detection right here.

We have, uh, Kensington Locks locked, uh, in the back. And then we have, uh, a bezel lock right here. You can log the bezel for security.

And then, uh, it has all the, uh, server security that we typically offer with our Proliance servers, the silicon route of trust I id platform certificates. And then we also offer the self encrypting drive for data security. And, uh, yes, and you get the optional, uh, rack mounting kit and wall mounting kit and, uh, for acoustics.

Um, it is definitely one of the quietest track server that we have. And the reason it's quiet because number one, um, we are using Ciena, which is very highly efficient. And, uh, number two, we are using the bigger heating and bigger fans.

And that's the reason why we went with Two U, because with two U, you get awesome airflow and, uh, when the airflow is good, uh, your fans are not loud. So that's why it's very quiet. Uh, the decibel range kind of, uh, ranges.

It, it kind of ranges from 35 to 55, depending on how you, um, populate this. And then, uh, as I mentioned, it can run from 23 degree Fahrenheit to, uh, 131 degree Fahrenheit, and it can also handle high shocks and vibrations. And then for management, we have ILO support, we have one view support, and we have the HP compute Ops management.

And then I, like I said, it is supported by a MD, uh, SB six four generation Ciena processors, and they range from eight course to 64 course. So very powerful. And then we have also qualified, oh, I forgot to mention, we just launched it.

So we also qualified the Nvidia L 40 s if you really wanna do, um, high, uh, inferencing at the edge. So we have the Nvidia L four and L four DS on this box. Um, I might have missed you covered, uh, um, airborne particles.

Yes. Um, how, how you address that? Can you repeat?

Oh, yeah, right here With the bezel, with the, Yes. The bezel has the dust filter inside. Yeah.

Is that a washable dust filter or is it Replaceable? It is, yeah. It is not a washable dust filter.

You actually, that's a good question. Um, you have to replace this dust filter every three to six months, uh, depending on how the environment is. We have a feature on ilo OneView and com where, um, on the 85 day of operation, the server will alert the customer that, hey, it, maybe it's time for you to check the filter, and if it's dirty, you need to replace it.

And we do offer replacement filter kits with this. They can purchase that. Yeah.

When's ILO seven coming? Hmm? When's ILO seven coming?

So that is a great question. We are on ILO six right now. We just launched it in September.

It's still a brand new product. We are going to have a refresh of this box soon, but it's still going to gen be Gen 11, which is ILO six. And the reason is that, um, a MD has done a great job in making sure the next generation processors are compatible with this system board.

So we don't have to change anything in the box. So the next generation will have the same box with the new processor and some additional cool features that we are working on Feature. Okay.

I think with that, I will hand it over to Gary, our HP Edge Compute Solutions product manager, and he's gonna talk about all the incredible solutions that we have. We're gonna show one example. So thanks everyone.

Um, so this, this slide, this used to be a IDC bar chart, but I figured bar chart is boring. So I took the data and change to this. So, but all these are benefits that Edge computing brings, or when you bring compute to the edge, these are the benefits that customers are seeing.

And I figured this, this shows a little bit better than just bar charts because I, in my view, they're all linked together, right? You, by bringing Compute to the Edge, you're improving data security because you're now processing everything generated at the edge. And by doing that, you're reducing connectivity cost, and you're getting faster, uh, decision points from your data, uh, creating better customer experience.

So those, that's why I believe this view shows a better, but there is that core reason of why bring compute to the Edge is you need to bring, you need to process that data generated where it's generated. You need to meet within that, what we call a real time window, because in traditionally, you're passing the data outside to your centralized location for processing or analytics. And once you get the data back, you're really outside of that real time window, right?

That data sort of becomes invalid. And, and that's why, um, bringing data to the edge is, uh, bring compute to the edge is now what everyone is doing. And that ties to a period is showing by 20 27, 20 5% of the server will be at the edge.

And this, this really shows all the industries that Edge computing can apply to, and how we build a common architecture like the DL 1 45 that meets all the challenges of these scenario, uh, these industries are facing for edge computing, but not just the common server architecture. But from a, your infrastructure, architecture perspective, as when we deploy a solution, uh, we want to create a solution blueprint that will apply to many, uh, industries at the same time. And also the systems and tools or management where, uh, Chris will later talk about com.

So we take retail, for example, right? Uh, customers, you have a retail store, you have IP cameras on site already, and then how do I turn that into something that I can deploy at the edge when I bring compute over? So our idea, the blueprint we're showing is you're, I'm gonna take 1 45, for example.

Uh, in this case, we have IP cameras realtime streaming to a one, DL 1 45 for realtime analytics. Um, we have a second DL 1 45 for all your day-to-day operations, right? So customers are saying, we want to start deploying ai.

We want to do computer vision at the edge. We want to analyze all the video streams, but we still have our traditional business operation applications. Uh, those will reside on a separate DL 1 45, or you can do it in a same system if, um, if it's a small site and you're not streaming a ton of cameras for analytics, right?

Uh, but the idea is to run your point of sale system, your man, uh, inventory management systems, your time core systems. Uh, but if you're a manufacturing site, you have your, um, uh, you can analyze your, uh, uh, like the, the defect, uh, detections. I kept thinking fraud detection, but that's FSI, which is also a use case for, um, for Edge.

And our idea is you are now processing everything on site, and you can pass on to, uh, your centralized data center for, uh, future training, right? Historical data for future training. And you have it, it, uh, IT team that can act, uh, manage your distributed edge site.

So if you are, um, we're in Texas and we use HEB, right? Let's say you have all the H-E-B-H-E-B sites are now doing, uh, this deployment, you're analyzing flow and analysis, uh, smart inventory management and things like that at the, in the store. Um, but you have your centralized, um, IT staff to manage the entire deployment.

So this blueprint, I'm gonna keep calling it Blueprint can apply to a manufacturing site, right? It can apply to a healthcare, uh, location, imaging center, uh, things like that. So how do we deliver it?

This really shows a lifecycle. Uh, the goal is to show a lifecycle, uh, management of the deployment. So up top, you have your edge infrastructure.

So this would be our DL 1 45 in this case, uh, management from the cloud. That's where our compute ops management comes in to stand up the entire environment to manage your hundreds and thousands of nodes, uh, distributed edge sites, um, and continue to manage the firmware and all the updates through that lifecycle of your, uh, environment. But if I dive a little bit deeper to just the infrastructure side, because I know Chris is gonna deep dive into the, uh, compute ops management for, um, for all your management piece, um, I know there was questions about the storage piece as well.

Um, and Electra, you know, when we talk about edge computing is not, you know, if the storage on the system is not enough, we always have that additional discussions about Electra. Do you need Electra on site? And also our Aruba networking, right?

Uh, it, we'll need networking capabilities. Uh, we have our Aruba networking, that's when we talk solutions. We bring those, uh, pan HPE products into, um, the discussion as well.

Um, and of course our, our hardware platform. But at the same time, we also partner with many, um, uh, industry pointers, like canonical, uh, HPE. We have our, uh, VME virtualized, uh, VM essentials, right?

Uh, Microsoft, uh, we have, uh, Azure, Azure local integrated system on the DL 1 45 ranchers qualify special cloud. We actually have this, uh, neat, uh, edge in a box so you can have special cloud installed from the factory, um, to, to the customer. Uh, but, Uh, I'm sorry, the, this is the slide before.

This was the first slide that I saw with GreenLake on after having a very good GreenLake smoothie. By the way, this morning. The, uh, the question is, uh, from a cloud management per perspective, I see the compute ops.

Can I do, uh, is this a target for my GreenLake workload? So as I'm thinking about my enterprise overall strategy for deploying apps, containers, et cetera, is this a potential target for, uh, a cloud architect, uh, trying to, uh, d determine where workloads should run, refresh of those workloads, AB testing, all et cetera, From the compute ops Ops, uh, from my GreenLake from, if I'm looking, if I look at this from a GreenLake perspective, I'm a GreenLake customer, and I need a endpoint to run my GreenLake, uh, uh, VMs or my GreenLake optimized work workloads. Is this a target for that, for that, uh, for that compute?

I believe so, yeah. Yeah. But right now we're calling the flex solution, right?

It's, it's not, it's a, it has to be a custom solution from GreenLake. So you guys are delivering the edge, the far edge devices in a GreenLake outcome, It can be a GreenLake deployment. Yeah, We do have a customer on, and they have edge device, h, HP s device.

It's of it. Yeah. That's Great.

It can be a GreenLake deployment. Good. Yeah.

Yeah. It'd be nice to put it in like a colo somewhere or Something. Yeah.

Well, But there's gonna be both, right? Yeah. Both.

Yeah. Yep. And on top of the audio OS layer, uh, we also partner with ai, uh, ISVs, uh, which I know we have an AI session later in the afternoon, but I want to call those out that we partner with clarify.

IU they rebranded vi dio. Um, I need to change the logo, but, and also StrataVision. So those are some of the ISVs that we're partnering with to deliver a full end-to-end solution that will meet, you know, retail customer needs, manufacturing customer needs, uh, for logistics for healthcare.

And I think with that, I'll pass it to Chris or computer ops management. Yeah, thank you. So as mentioned earlier, Chris Bradley, uh, I lead the tech enablement team.

And when I start thinking about edge deployments of servers, um, I, I think about how the pendulum has swung. It used to be that, uh, a lot of organizations would put equipment out the edge, and then there was a, there seemed to be a trend of, hold on, that's really hard to manage and maintain, and I'm tired of putting technicians on planes or having essentially guys driving across the country to go work on, on equipment and or dispatching a whole team to go stand up. And I know it's a, a use case that we've talked about a lot, but stand up a new retail shop, um, or to tear it down for that matter.

Um, so this idea of deploying servers at the edge ended up being a little bit complex. Um, and if you just use one of the examples that Gary talked about, um, let's use Azure local little three node cluster running on the deal, 1 45 Gen 11. Um, and I only have 10 office locations or branch locations, that's 30 servers that I, as the infrastructure guy have to go manage.

And if I don't have a consistent tool set to do it, and it doesn't necessarily look like what I operate my data center with, that could lead itself to some abandoned infrastructure that it was set up at one point. We don't really do a whole lot of operational maintenance on it, um, because the tool set's different, it's out of sight outta mind. Uh, so HPE has been on this trajectory, uh, for a number of years now to support, um, our, our customers with a unified experience for management, whether it's in the data center at the near edge or the far edge, um, with HPE compute ops management.

Now, it was mentioned earlier, um, HPE GreenLake Compute Ops Management is an application within GreenLake. So whether it's deployed as a GreenLake offering or managed, um, it is still through the GreenLake platform that our administrators will manage the edge as well as the data center. So GreenLake, um, and specifically compute ops management has again, been on this trajectory of trying to accelerate the usability of our systems.

Um, the, you know, trying to regain some of the IT administrators time, uh, for simple tasks like, uh, firmware and driver maintenance, just the care and feeding of a server, uh, we have to be able to do it from afar. So the administrator, usually the IT administrators that are working on edge systems are also the ones responsible for the data center. So they're centrally located.

Um, and we can save a ton of time and money by not putting people on planes or in trucks to go do those maintenance operations. Um, so in integrated with every platform, you heard it from Darren, uh, it's expected that this deal, 1 45 has IO embedded. And in fact, uh, future presentation, we'll talk about how everything that we ship has ILO embedded in everything.

So that silicon-based root of trust is embedded in a three HPE ProLiant we ship. That is the foundation that enables the remote management experience. Um, for compute ops management, it is the telemetry provider for compute ops management or comp that allows us to aggregate everything, uh, at a higher level through software.

We don't have to manage ourselves. Imagine, and frankly, I grew up in the industry managing my management tool, um, having to do firmware or having to do software and security patches of my management tool only so that I can manage the tool in which it is managing, if that makes sense. So exception.

Yeah. I got a little laugh from the group. It's like, yeah, why, why manage my management tool?

Um, so if I'm Understanding mm-hmm. Please, Um, the, the relevance of this to edge is that, um, a DL 1 45 based edge deployment has ILO in it because does it has ILO in it, and I assume you can do, you know, remote updates of ILO, of course, right? GreenLake probably handles that, but because it has it in it wherever you've put it, um, assuming connectivity, which is really shouldn't assume at the edge, but let's just roll with it.

Um, your GreenLake cloud-based GreenLake, wherever you are, or your operators in the world, can manage that server or that cluster in the same way they're managing on-prem. Thanks to this sort of, I won't, I don't wanna use the word agent, but I'm gonna use it anyway. I am sorry.

Uh, be, uh, the IILO. Yes. Um, is there, so there's less buts on planes, um, because, uh, you, for those sorts of updates, um, you can, you can do it reliably, remotely.

Are there particular elements of compute ops management that, um, take account of connectivity issues, availability issues, latencies, that sort of thing that are typical in edge as well, just so that that automation can, can take place in continue despite a different operating environment? There are a number of things we can do to mitigate some of those edge challenges. Things like latency, uh, intermittent connectivity, uh, that are physical security also can be, yeah, physical security.

So of course, ilo, again, being the, I'm gonna start with the physical security piece, ilo, of course being the telemetry. Um, for things like chassis intrusion or device removal or, um, system attestation, every time it boots to ensure that, uh, firmware that is loaded is still legitimately from HPE and signed. Um, the concept of grouping allows us to group things, uh, not only based on, let's say application, but geography, uh, located, uh, depending on the deployment.

And I am getting into a, a future session that you'll hear about that's dedicated to compute ops management. But think about a local secured gateway that if I plant basically a virtual machine in the edge location, it becomes a broker for the devices that is managing, um, for things like communication back to HPE GreenLake, um, as well as staging of packages for things like firmware updates. So envision a world where connectivity to one particular site on the west coast of the United States, for example, just is for whatever reason, intermittent, if the secure gateway contains all of the firmware and driver packages that I'm gonna update at a future time, you're less likely to encounter a severed connection back to GreenLake while that firmware update is being executed.

Yeah. If those dominoes were to fall, um, in that particular order, you'd still end up with a su successful update because the contents of what is going to be updated is already local at that site. Yeah.

Yeah. I think, I think to, to guy's point though, right? Mm-hmm.

Or packaging perspective. Yes. Right.

If I'm thinking about edge, right? I don't wanna just think about compute, I wanna think about the network, the wifi and all the things that HP sells in you. You guys sell it all and you deploy it all.

Right? And I would love, because from a compute ops, I want an edge ops management platform, right? Are you guys getting it into that where you have one platform to manage network compute, perhaps?

Uh, the security, uh, across the board? I think our solution is HPE GreenLake, because within IT, we have Aruba Central, we have data services cloud console, we have compute ops management, we have, um, uh, the ops ramp. Um, there are a number of tools within the platform.

Compute ops management's only one of them, um, that enables that portfolio level management, whether it's wireless at the edge, whether it's data center networking, whether it's compute in the data center or at the edge. Uh, and the same thing with storage. So I think that is the platform that HPE provides.

Okay. Well, Here's a twist on that one. Sure.

What about redfish? Are you, are you leveraging tools like that, that are industry standard to cross pollinate across these different, you know, silos? Are you enabling toolkits to, uh, turnkey manifest and turnkey execution?

So day zero, we have a bunch of pre-program turn it on, boom, you're done. You know, are, are those things in the thought process of, of edge consumption today? Are there still, you know, wishful thinking?

I think there are a number of things that are, uh, already available as well as in the works, um, from a, let's say day zero, uh, operation. Mm-hmm. You'll see in the demonstration a little bit later that there can be policies set up within compute ops management, uh, in the concept of groups and adding new devices to groups and make magic happen.

Uh, so policies are invoked on, on, uh, you know, bare metal servers all the way to an operating system deployment if necessary. I mean, you can, you, because you have the ability, you're, you're the vendor, you're the manufacturer, you can standardize any slot. Mm-hmm.

Any config, any bios you can, you can make, like I said, magic happen, right? Yeah. So, and I, I'm going to, uh, just iterate on, uh, the comment around redfish.

Yeah. Um, whether it's platform enablement at the individual server or, uh, leveraging APIs from GreenLake or specifically compute ops management, we have that ecosystem as well. Um, one of my favorite innovations as of late has been a, um, PowerShell command bundle for compute ops management.

So I can leverage, um, my keyboarding skills and or the code at once, deploy many, um, to, uh, to pull the APIs of compute ops management and be the puppet master, uh, for deployment and management of my tools. Yeah. 'cause we're talking about all these different silos.

Is there composability? Can you make that, you know, 'cause redfish is obviously day zero, it's the hard work, but once you are past that, you have the, you know, what's your hypervisor of choice? What's your, is it VME?

Is it, is it Windows? What do, what's the Edge marketplace looking at right now? And are, are the, you guys changing based on cots, based on different, different cots and different type of data?

I mean, help me understand. Yeah. The, the edge has been evolving very quickly.

Um, obviously we've got great partnerships with a number of ISVs that Gary was talking about, um, as well as our own solutions being, uh, H-P-E-V-M-E, our virtual machine essentials. Um, and so, you know, choice is there for customers, uh, and we, uh, fully embrace those choices, uh, whether it's Microsoft or HPE or Red Hat, or suse, or VMware, whomever. Um, so we embrace that choice, uh, and do our due diligence to make sure that it operates as expected on HPE ProLiant.

Yeah. Have you been seeing a lot of any customer, uh, uh, requests for support of things like redfish or, or those open standards? Um, if we do, they don't necessarily maybe make my radar, because the answer is absolutely yes.

We are a hundred percent redfish compliant. Um, you know, HHPE was a founding member, um, uh, pushing for that standardization that is called Redfish. Um, and, uh, so yeah, that's, that's a very intrinsic to every machine that we, or or every IO code set that we have.

I did, I did it like a two 1200 server build for redfish. So there's customer need, there's Customer, there is customer need. And that redfish enablement, uh, probably started back with Gen 10 and has continued through plus 11 and now Gen 12 as well.

So kind to land the plane here, uh, on the edge of deployments. Uh, while we've focused a lot on the DL 1 45, I do want to reinforce that we do have solutions purpose built, uh, for given scenarios. And, and just essentially in general, you can see this ultra compact, um, has been around for quite a while, what we call the micro server.

Um, so very tiny package. Um, typically an Intel based platform, no GPU support, but great for lightweight point of sale, uh, SMB file share, let's say small medium businesses take a file shares, printer servers, stuff like that, um, to a, uh, rack mounts as well as tower versions of, uh, some of the Intel based machines. We've talked quite a bit about the DL 1 45, um, as it is one of the newer kids, uh, to the portfolio.

Um, as well as some platforms that are really, um, ruggedized, um, maybe more telco focused, DL one tens, uh, as an example, but another one of those machines that interestingly are front powered and front cabled, um, so that we're optimizing the cooling and, uh, for those type of deployments. And then of course, there are some environments that, uh, still need a, um, uh, to follow their corporate standards and wish to deploy DL 360 and three eighties at the Edge. Certainly applicable products for those.

Uh, but maybe that's a one size that doesn't fit all, which is why we have some of the other flavors in the portfolio. I just love to see a plug and play stateless edge OS level. Once you've provisioned it for redfish and then you just plug in what you want and you're, uh, you know, can consume and start doing real data analysis, that's what you wanna do, right?

You want get it faster to that Point, right? Yeah. That, that is a lot of, of what we want to do, uh, is be able to provide an edge location with a machine, with something here on the, uh, on the portfolio and with zero experience, other than take the blue cable and put it into the RJ 45 slot, um, and then turn it on or just plug it in, uh, let automation take the rest.

Uh, so that's absolutely where we think the edge is headed, and we're doing a lot of work in that space. Thank you.