 We're back in Las Vegas. The Venetian Conference Center, Dave Vellante and Rob Streche, Lisa Martin's also in the house. She's taking a little break right now, but this is our coverage, the CUBE's coverage of HPE Discover 2023. We extract the signal from the noise. Unified platform, GreenLake is all the talk. LLMs is a service. Mohan Rajagopalan is here. He's the Vice President and General Manager for HPE's Esmeral software. And Brian Thompson is the Vice President of GreenLake Cloud Product Management. Also at HPE, of course, Brian. Good to see you, Mohan. Thank you for coming on. Hey guys. Thanks. Thanks for hosting us. Yeah, you're welcome. So obviously GreenLake, we've watched the ascension and maturity of GreenLake. It's like, I feel like it's finally there. We saw Fidelma this morning lay out the platform. Nice story, clean, I get it. Okay, you know, boom, now it's go. So give us the update. What's new, what's exciting to you, Brian? Yeah, for me, I think you really hit on it. Even this time last year, I think we were introducing the GreenLake Cloud Platform and this journey to one, we're seeing more and more where all of our offerings, GreenLake itself being this large number of breadth of offerings, now converging and landing on the platform, how do we start to give more of that unified experience to discover and consume these services? So it's exciting to see that really happening. There's a lot of work that's taking it to do this. And for me, from my side, I'm really on the private cloud and where we're seeing that start to grow and adopt in the momentum behind that, part of that full experience. So it's been neat to see where we were kind of planting the seeds last year and now really seeing them take off. You know, it's funny, I mean, probably about, I want to say 10 years ago, and we wrote defined true private cloud, we called it. And we were, you know, we said, okay, this is what it's going to look like. And it took a while to get there. You know, you could take some of the HCI platforms and say, okay, that's kind of there, but they're very limited in terms of the other services that they had. And now that vision that was painted 10 years ago has really manifested itself in the form of GreenLake and other types of services. But, Mon, I wonder if we could explain to the audience Esmeral. It's a lot of things to a lot of people and there's a lot of capabilities in there, but how should we think about what it is and how it interacts with GreenLake? So it's a great question and thank you for the question. Esmeral has meant a lot of things for a lot of people. You've gone through several evolutions. Esmeral, I think the easiest way to think about it is, we are the software business unit under the broader GreenLake umbrella, right? Esmeral started off as a bunch of acquisitions. We had acquired startups like BlueData and MapR Lampool. And you know, I joined HPE about a year ago. And when I started at HPE, Esmeral was really a collection of really cool technology all the way from HPE's own container platform technology through to data technologies, through to like, you know, ML ops and data ops kind of use cases solutions. Over the last year, we've gone through a big transformation where we started to be very focused based on two pieces of input. One was customer feedback. We've talked to over a hundred customers, specifically to try and understand where can HPE and GreenLake particularly help our customers in their software journeys, right? Second was also talking to our salespeople. You think about HPE, HPE is known for support and services in the infrastructure world. If you think about our sellers, like you know, it's really hard for a seller to go out and sell some very sophisticated software solutions. So taking these two perspectives into mind, last year we focused the Esmeral portfolio to focus on two value propositions. One is how do we make it easy for our customers to manage data in a variety of formats? So think about objects, files, streams. Now we're doing iceberg. So think about snowflake, data bricks type use cases coming to on-prem as well as vectors, graphs, et cetera in a single pane of glass. Along with this, we're also focused on bringing the latest, greatest open source technologies in a fully managed offering to our customers on-prem in the cloud, as well as in hybrid setups. So I want to follow up on that. So snowflake on-prem through iceberg. So essentially through a materialized view or how does that work? Fantastic. So we've basically extended our data fabric technology to include native iceberg tables. So we can support both external tables as well as internal implementations of tables. Now what differentiates our implementation from some of the other hybrid cloud solutions is native iceberg implementation. So as a result, we don't create copies of data. We don't have to marshal data into SC and then say try to compress it, hide it, et cetera. We simply are able to select particular key values that basically need to be processed and transport them where it needs to be consumed. So more looking at it from a data mesh perspective versus then doing almost a, I'd say ETL, not ETL but ELT from that. Exactly. So I think the vision, so we've had lots of debates around the choice of named data fabric. I took an opinion at the stand saying I think the future of all data platforms is going to be a combination of a mesh, is going to be a combination of lake houses, et cetera. And we think fabric is an amorphous enough term that in some sense represents our future-looking vision. So when we think about data fabric, it's really a collection of distributed data sources that are in some sense interoperating with one another. When you talk about tables, I think it's a great use case where I think a lot of financial customers as far as adopting technologies like Snowflake, you want to make it easy for them to be able to access data that's been fed into the fabric. We're also seeing a lot of customers in the AIML world that care about object stores, et cetera. We want them to be able to look at the same pieces of data without creating replicas and copies just through higher level interfaces of sorts. That seems like it would be huge for the cloud side as well and how they play together. Yeah, we used the analogy recently of peanut butter and jelly with our two groups going together. One of the key things we've had is things like data fabric. Being inherently hybrid cloud enabling, how do I deploy this in private or public cloud, single global namespace that spans? But with things like our HP GreenLake for private cloud enterprise, it's that natural extension. How do I have this as the substrate that I land those data services on, that fabric on, that can now extend into that same hybrid experience? So we're seeing a lot of traction in kind of delivering this as a tightly integrated solution together. So you basically describe it as HP's software portfolio and it might okay to say it's also sort of the data management piece is a big chunk of that portfolio which is critical for your customers. Absolutely. So when we think about our enterprise customers, we want to provide two foundational capabilities for our customers. One is data management. The second is a set of analytics tools to process the data that's in the system. And that's kind of been like what our core focus has been for the last year or so. We think the message is resonating well. I think it's a simple enough message. We're not trying to do everything for everybody. We're trying to be very focused on where we think we can complement technologies like Ryan's private cloud enterprise and in some sense offer our customers that value-steal all the way from the infrastructure layer to the core foundational software assets that they may want to build on. So you're providing optionality. You mentioned Snowflake before and I'm just choosing them but there's probably others that you can work with. Anybody can use an iceberg table, right? So in that sense it's open but you're not trying to specifically compete with those data platforms. I mean there's overlap but are you trying to be the database of choice? No, I think it really comes on. I think there are so many tools and vendors of choice available today. I think it really comes down to giving our customers a choice for the solutions that they want to build, right? And this is where I think HP differentiates itself. I mean historically we were not known to be the stewards of software especially in the cloud and cloud native spaces but we're taking a very interesting stance here today where we have an opportunity to use the HP brand and presence to bring together the best of great open source technologies with open interfaces and open standards such that our customers get the best range of choices in terms of how they want to stage the data, what kind of tools they want to use and how they want portability, right? We're big believers of as we start seeing hybrid and we can talk to some customer stories here where it's really about giving our customers the choices to place the data and compute where they want to across different types of operationalization. And given them options on different types of platforms I would think whether it's OLAP or OLTP, Vector now, Graph databases, and these the types of things that it's streaming, are these the types of workloads that you aspire to support or can support? Absolutely, I think the answer is yes to all of them today. In the past we used to spend a lot of time with customers to try and figure out what kind of workloads they want to run. What we've noticed over the last year with our open interfaces strategy is our customers have now leapfrogged us. We have customers that are already doing vectors and graphs on data fabric before we natively supported it. And this is actually creating a great feedback loop for us because now we get a chance to actually move away from being a transactional vendor to becoming a strategic partner. We spend a lot of time with our customers learning about their use cases and trying to see how that influences our roadmap. So how does that work? Do I bring my own platform to GreenLake or can I spin it up as part of a console or? I think there's a couple layers to it. It starts with, as we kind of mentioned, customer choice. So even in those foundational IaaS components, we've done things like even within the Kubernetes layer, we think about our private cloud experience. We started with just providing the Esmeral Kubernetes runtime. We've enabled customers to choose Amazon EKS anywhere, read that open shift, continue to add those flavors. But then even as we move up into this, I think Mohan and the team with Esmeral have really done a good job of building frameworks that allow you to choose from a portfolio of solutions but even snap in your own, right? Where we can handle things like security and identity and the other things that would take to integrate that but without being overly prescriptive and therefore limiting on customer choice and what are they trying to achieve? So I think I've totally blown all the things we were supposed to talk about but let's stay on this. So we had Uber on last week and we've been using Uber as sort of an example of the future. When you think about the Uber app, it's the first app that actually sort of took a physical ecosystem and brought it together digitally. People places things, drivers, riders, ETAs, et cetera and turned those into coherent data elements. Do you, in real time, quite remarkable actually, do you see a future where your customers are going to sort of drag you into that world of real time data coherence and basically building a digital twin of their business? I think it's already happening. So if you saw Vishal Lal's keynote today, we had BMW on stage where BMW talked about the current use cases where they're using a combination of GreenLake as well as Esmeral and what was very fascinating for me was they're talking about doing things like autonomous driving but also looking at security systems where in real time they're monitoring and they'll be able to do predictive analytics on when failures could happen in a car and proactively either notify the driver or course correct such that the problem is averted before it can actually ever show up, right? So it's not about how we help our customers build the future, they're already building the future and we're learning about it as we speak today. Is it, do you aspire to be a development platform ultimately where people can build data apps on top? So when you look at Esmeral specifically, our target audience is the Fortune, it's largely enterprise development teams in Fortune 50 to Fortune 2000. That's where we see a sweet spot. I mean, what has really driven us to this point really is we have a very strong presence in, HP has a very strong presence in the enterprise community. We are seeing a lot of traction and interest in tools like HP's private cloud offering in the enterprise community. We want Esmeral to be the foundational layer of software capabilities so our customers can develop solutions in-house, right? Without any restrictions to choice of technology or format of data. I think one thing that you actually hit on, Brian, a minute ago was just the fact that you have these different partners, the EKS Anywhere or Red Hat. I saw it before I got here that Red Hat was the technology partner of the year for 2023 and I think that means last year for you guys but I can't remember when, it was in January or something like that. How are you seeing the adoption of Kubernetes and containers on cloud, right? The private cloud right now. So it's been a critical component of each one and that's why I think we've seen that embracing of the cloud itself. They're not just looking for, hey, update my VMware environment. I'm looking for that cloud experience which is providing me additional cloud primitives like containers. I think we're seeing with that adoption I have existing workloads that I'm opinionated that I'm going to consume. That's why they're let me specify what Kubernetes runtime and the tool chain that I'm using. In other cases, it's how do I consume this as a service? Meaning let's take away the complexity of hand rolling these clusters and managing myself. How do I consume it as a fabric service within that private cloud? So it's been an enabler for customers on various points along that spectrum. They're either highly sophisticated, opinionated and drive their own versus I have developers that want to consume this technology I don't want to operate it. How can I consume it as a service? But it's been an integral component of every conversation because whether they're there now, it's where they're going. So they want to make sure it's a part of that journey. You guys were chatting about ELT before. I extract load transform as opposed to ETL, right? That's kind of how we did it in big data, right? Just throw it in there and we'll figure it out later which kind of created a mess but then over time tooling became better to be able to do that. But if you have all these different data types and you've got to make them coherent, you can do some heavy lifting but is there the potential for a semantic layer that emerges that actually serves as that translator for all these different data types? Great question, I think the answer is yes. But it's not one semantic layer, it's a continuum because it really depends on the workload, right? I'll give you some concrete examples. So we have very large class of customers who basically focus on the traditional ELTTL type of database workloads, right? This is where, like out of the box today in Esmeral, we support Hive Metastore and it's a very structured catalog of sorts. If you look at some of our more forward-looking customers who are doing AI ML type use cases, there is no context of a catalog or structured data that can be like, you know, from which metadata can be extracted, right? This is where we still see, in terms of repeatability, we see fluidity of context being extracted, right? And as we transition from files and objects to graphs and vectors, we see, we're going to start seeing more and more of these metadata formats come out, which will be very workload specific. So AI and ML workloads may have vector representations that form the metadata layer, whereas the traditional database workloads may have iceberg tables or like Hive Metastores come to the next layer, right? From an HB Esmeral point of view, we don't want to have a strong point of view saying, this is the way to do things. It's really about following the puck where the business is headed to and making sure we give our customers options. Okay, so let the platform, the data platform, people fight it out, snowflake data bricks, and you'll support whatever. You don't really care, right? You just want to have the best infrastructure as possible to manage that. I think we want to be the Switzerland in terms of technology, right? We want to support everybody. We want to, in some sense, be a facilitator. At this point, I think it is really, like the market is so crowded and there are so many strong opinions. I think it's really about giving our customers a voice and a support system where they can basically start building true solutions. But it's interesting because you have to sort of thread the needle on differentiation. You don't want to be pure commodity hardware, and so you've made some acquisitions, you've done some organic development, a GreenLake IP, so that you can differentiate and have a higher margin business. At the same time, you've got to be open and people want that commodity-like experience in terms of simplicity of buying, at least. I think it's a way, and you're right. It's really about how do we thread the needle? So this is where we're choosing not to create bespoke IP and create walls, which basically lock our customers in-house. However, if you look at a combination of PC and Esmeral, we are able to do things like zero trust security out of the box, right? We think productivity, security, et cetera, are bigger differentiators instead of trying to create a walled garden of sorts. What do you think about that? I was going to add, if you recall this, we recently added our customer advisory board a couple months ago, and the statement was made as we were kind of showcasing where we've gone with PC and running Esmeral on top, and that framework that I kind of mentioned, where it's not overly prescribed on, these are the tools that you can use, rather than, here's a framework you can step and bring your own. And the statement was made saying, I don't know what my team's going to ask for six months from now, so the framework and the ability to easily integrate that, leverage security, and the other things that I care about, that's what was important to them. We're like, okay, we're on the right track. This is going to be valuable. And I think you hit on a very interesting topic, because I think the people who buy that data layer are not typically the people who buy the cloud layer as well. How is that working out with like, you're seeing all these users and these user groups. I think it's a journey. I think like, you know, it's only recently, even we got the awareness around the difference in persona, where the persona who's buying the IT stack is different to the persona who's buying the analytics stack. But I think we're starting to see convergence. And I think our big differentiator is the set of options. I truly think today HP provides the maximum breadth and depth of options. It's not just about taking commodity software and hardware and bundling it together. We've actually spent time to differentiate our offerings, right? If you think about ease of use, if you think about predictable costs, if you think about the overall experience, I think we are in some sense unmatched in the market today. Yeah, I mean, that's that cloud operating model, right? And that's how you get people across there is they don't have to, you know, they're not swiping the credit card every time they go in and do something, or maybe they are, depending on how I guess they would acquire it, right? But they're, you know, if they're getting it as a service, you know, they'll buy it by the drip from that perspective. But yeah, that makes total sense. And I think that's where, you know, I think it was Jim Jackson, we had an analyst group meeting with him. I think that was actually yesterday. And he was talking about how marketing is actually going out and talking to data scientists, for instance. That would seem like yet another one that is interested in what Esmerell and cloud can do for them. And, you know, obviously the LLM announcement as well. How do those all tie together? Kind of those, is that where you're seeing Esmerell kind of being the data, kind of the data fabric for getting to those LLMs and- You know, if you think about it from an HP point of view, again, it comes back to giving our customers the maximum set of options. I think we are personally very excited about the GreenLake for LLM capability because it's really bringing our strength and supercomputers to a large class of like enterprise customers that could typically not afford a supercomputer or would not invest in one. If you think about Esmerell, Esmerell provides foundational data and like, you know, open source tooling capabilities. If you look at our machine learning development environment or the package acquisition, we are also building our bespoke tools which are in some sense catered for these high performance execution engines So again, coming back to the theme around HP providing the best possible breadth and depth of options. I think it all just naturally comes together as a big jigsaw puzzle. It's very helpful in describing what the strategy is and the differentiation and the sort of long-term view. It's not like you have to accurately predict exactly what the next big thing is down the road. You just got to be there to support it, whatever it is. So, appreciate you guys coming on. Yeah. Thank you. All right, keep it right there. Up next, we're going to talk about the future of AI and some of the cool things that society can expect and how to make it responsible. Rob Streche, Dave Vellante, Lisa Martin. You're watching theCUBE's coverage of HPE Discover 2023 from Las Vegas, day two. We'll be right back.