 Okay, now we're going to dig deeper into HPE Esmeral and try to better understand how it's going to impact customers. And with me to do that are Robert Christensen, who's the Vice President of Strategy in the Office of the CTO and Kumar Srikanti, who's the Chief Technology Officer and Head of Software, both, of course, with Hewlett Packard Enterprise. Gentlemen, welcome to the program. Thanks for coming on. Good seeing you there. Thanks for having us. Always great to see you guys. So Esmeral, kind of an interesting name, catchy name. But Kumar, what exactly is HPE Esmeral? Yeah, it's indeed a catchy name. Our branding team done a fantastic job. I believe it's actually a derivative from Esmeralda. It's a Spanish for Emerald. Apparently it's supposed to have some very mythical powers. And they derived Esmeral from there. And we all actually initially, when we heard it, it was interesting. So Esmeral was our effort to take all the software, the platform tools that HPE has, and provide this modern operating platform to the customers and put it on the one brand. So it has a modern container platform. It has persistent storage, distributed data fabric. It has InfoSight as many of our customers familiar with. So think of it as our container platform offering for modernization and digitization for the customers. Yeah, it's interesting you talk about platform. So it's not, a lot of times people think product, but you're positioning it as a platform. So it has a broader implication. That's very true. So as the customers are thinking of digitization, modernization, containers and microservices, as you know, has become the stable hold. So it's actually a container orchestration platform. It offers open source provenance as well as a persistent storage voltage. So by the way, Esmeral, I think Emerald in Spain, I think in the culture it also has immunity powers as well. So immunity from lock in and all those other terrible diseases maybe helps us with COVID too. Robert, when you talk to customers, what problems do you probe for that Esmeral can do a good job solving? Yeah, that's a really great question because a lot of times they don't even know what it is that they're trying to solve for other than just a very narrow use case but the idea here is that to give them a platform by which they can bridge both the public and private environment for what they do in application development, specifically in the data side. So when they're looking to bring containerization, which originally got started on the public cloud and has moved its way, I should say, become popular in the public cloud and has moved its way on premises now. Esmeral really opens the door to three fundamental things like how do I maintain an open architecture like you're referring to to some low or no lock in of my applications. And number two, how do I gain a data fabric or a data consistency of accessing a data so I don't have to rewrite those applications when I do move them around? And then lastly, where everybody is heading out, the real value is in the AI ML initiatives that companies are really bringing that value of their data and locking that data at where the data is being generated and stored. And so the Esmeral platform is those multiple pieces that Kumar was talking about stacked together that deliver those solutions for the client. So Kumar, what's the, you know, how does it work? What's the sort of IP or the secret sauce behind it all? What makes HPE different? Continuing our theme of physical force around Esmeral, it's a modern platform for optimizing the data engines and workloads. That I think is, I would say there are three unique characteristics of this platform. Number one is that actually provides you both an ability to run stateful and stateless workloads under the same platform. And number two is, as we were thinking about, unlike if another Kubernetes open source, it actually gives you all open source Kubernetes as well as an orchestration behind it so you can actually do it. You can provide this hybrid thing that Robert was talking about. And then actually we built the workflows into it. That's for example, we have actually announced along with the Esmeral ML Ops platform that customers can actually do the workflow management around specifically data engines and workloads. So the magic is if you want to see the secrets of it, all the efforts that have been gone into some of the IP acquisitions that HPE has done over the years. We said we blew data, map bar and the nimble info side. All these pieces are coming together and providing a modern digitization platform for the customers. So these pieces, they all have a little bit of machine intelligence in them. People, we used to think of AI as the sort of separate thing. I mean, the same thing with containers, right? But now it's getting embedded into the stack. What is the role of machine intelligence or machine learning in Esmeral? Oh, I would take a step back and say, you know this very well, they're the customers data, amount of data that is being generated 95% or 98% of the data is machine generated and it has a serious amount of gravity and it is sitting at the edge. And we are the only one that have edge to the cloud data fabric that is built into it. So the number one is that we are bringing computer or a cloud to the data, they're taking the data to the cloud, right? So if you will, right? It's a cloud like experience that provides the customer. AI is not much value to us if we don't harness the data. So I said this in one of the blog posts we have gone from collecting the data era to the finding insights into the data, right? So that people have used all sorts of analysis that we are to find data is the new model. So the AI and the data and then now your applications have to be modernized. And nobody wants to write an application in a non-microservices fashion because you want to build the modernization. So if you bring these three things, I want, I have a data gravity, I have lots of data. I had to build an AI applications and I want to have an idea. Those three things I think we bring together to the customers. So Robert, let's stay on customers for a minute. I mean, I want to understand the business impact, the business case, I mean, it's, why should all the, you know the cloud developers have all the fun? You mentioned it, you're bridging the cloud and on-prem, they talk about when you talk to customers of what they are seeing is the business impact what's the real drivers for them? That's a great question. Cause at the end of the day, I think the recent survey that was that cost and performance is still the number one requirement for this real close second is agility the speed of which they want to move. And so those two are the top of mind every time but the thing we find in Esmerelle which is so impactful is that nobody brings together the silicon, the hardware, the platform and all that stacked together work and combined like Esmerelle does with the platforms that we have. And specifically, you know when we start getting 90, 92, 93% utilization out of AI ML workloads on very expensive hardware it really, really is a competitive advantage over a public cloud offering which does not offer those kinds of services and the cost models are so significantly different. So we do that by collapsing the stack we take out as much intellectual property excuse me, as much software pieces that are necessary so we are closest to the silicon closest to the applications bring it to the hardware itself meaning that we can interleave the applications meaning that you can get to true multi-tenancy on a particular platform that allows you to deliver a cost optimized solution. So when you talk about the money side, absolutely there's just nothing out there and then on the second side which is agility. One of the things that we know as today is that applications need to be built in pipelines, right? This is something that's been established now for quite some time. Now that's really making its way on premises and what Kumar was talking about was how do we modernize? How do we do that? Well, there's going to be some that you want to break into microservices and containers and there's some that you don't. Now, the ones that they're going to do that they're going to get that speed and motion, et cetera out of the gate and they can put that on premises which is relatively new these days to the on premises world. So we think both won't be the advantage. Okay, I want to unpack that a little bit. So the cost is clearly, really 90 plus percent utilization. I mean, Kumar, you know, even a pre-virtualization, we know what it was like even with virtualization, you never really got that high. I mean, people would talk about it but are you really able to sustain that in real world workloads? Yeah, I think when you make your exchangeable currency into smaller pieces, you can insert them into many areas we have one customer who's running 18 containers on a single server and each of those containers, as you know, early days of data, you actually modernize what we consider we run containers of micro VMs. So if you actually build these microservices and you have all anti-affinity rules and you have provisioning formulas all correctly, you can pack, impact these things extremely well. And we have seen this, again, it's not a guarantee, it all depends on your application and you're, I mean, as an engineer, we want to always understand how these carriers work but it is a very modern utilization of the platform with the data. And once you know where the data is and then it becomes very easy to match those two. Now, the other piece of the value proposition that I heard Robert is it's basically an integrated stack. So I don't have to cobble together a bunch of open source components. It's there, there's legal implications, there's obviously performance implications. I would imagine that resonates as particularly with the enterprise buyer because they have the time to do all this integration. That's a very good point. So there is an interesting question that enterprise, they want to have an open source so there is no lock in but they also need help to implement and deploy and manage it because they don't have the expertise. And we all know that K8S has actually brought that API, the pass layer standardization. So what we have done is we have given the open source and you write to the Kubernetes API but at the same time orchestration, persistent storage, the data fabric, the AI algorithms, all of them are bolted into it. And on the top of that, it's available both as a licensed software and run on prem and the same software runs on the green lake. So you can actually pay as you go and we run it for them in a call or in their own data center. Oh, good. That was one of my latter questions. So I can get this as a service, pay by the drink. Essentially, I don't have to install a bunch of stuff on prem and pay a perpetual license if I choose. It's a container as a service and MLAOPS is a service in the last discover and then now it's gone production. So both MLAOPS is available. You can run it on prem on the top of Esmeral container platform or you can run it on the green lake. Robert, are there any specific use case patterns that you see emerging amongst customers? Yeah, yeah, absolutely. So there's a couple of them. So we have a really nice relationship that we see with any of the Splunk operators that were out there today, right? So Splunk containerized their operator. That operator is the number one operator, for example, for Splunk in the IT operation side or notifications as well as on the security operation side. So we found that that runs highly effective on top of Esmeral, on top of our platforms that we just talked about with it, Kumar just talked about. But I want to also give a little bit of backgrounds to that same operator platform. The way that the Esmeral platform has done is that we've been able to make highly active active called HA availability at nine, excuse me, at five nines for that same Splunk operator on premises on the Kubernetes open source, which is, in far as I'm concerned, very, very high end computer science work, if you understand how difficult that is. That's number one. Number two, is you'll see Spark, just to Spark workloads as a whole, all right? Nobody handles Spark workloads like we do. So we put a container around them and we put them inside the pipeline of moving people through that basic ML AI pipeline of getting a model through its system, through it's trained and then actually deployed to our ML ops pipeline. This is a key fundamental for delivering value in the data space as well. And then lastly, this is really important. When you think about the data fabric that we offer, the data fabric itself doesn't necessarily have to be bolted with the container platform, the container, the actual data fabric itself can be deployed underneath a number of our, for competitive platforms who don't handle data well. We know that, we know that they don't handle it very well at all. And we get lots and lots of calls for people to say, hey, can you take your Esmeralda data fabric and solve my large scale, highly challenging data problems? We say, yeah, and then when you're ready for a real world, full time enterprise ready container platform, we'd be happy to prove that. So you're saying if I'm referring correctly, one of the values is you're simplifying that whole data pipeline and the whole data science, science project, pun intended, I guess. Yeah, that's true. So, where does a customer start? I mean, what are the engagements like? What's the starting point? HP is probably one of the most trusted enterprise supplier for many, many years. And we have a phenomenal workforce of the both, our point next is one of the leading world leading support organizations. There are many places to start with, right? One is obviously all these services are available on the green light as we just talked about and they can start on a pay-as-you-go basis. We have many customers that actually somewhat grandfathered from the early days of the data and map are, and they're already running and they actually improvise on when, as they move into their next generation modernization. You can start with simple azimuthal container platform with a storage computer service operation and can implement as little as 10 nodes and to start working. And finally, there is a big company like HPE as an enterprise company with the point next services. It's very easy for the customers to be able to get that support on a day to operations. Thank you for watching everybody. This is Dave Vellante for theCUBE. Keep it right there for more great content from Esmeralda.