 Welcome back everyone to theCUBE's live coverage in Chicago. I'm John Furrier, your host here on the ground. KubeCon, CloudNativeCon, part of the CNCF. theCUBE's been to every single CNCF KubeCon. Never missed one in Europe, never went to North America because all the actions here, all the authentic builders are here, the big vendors are here, the entrepreneurs are here, all the top open source projects, and all the innovations here. This next segment is with Platform9, Kube alumni, Madhura Mikhaski, VP of Product Platform9 is here. She's been on before, but now they got something new because obviously with Kubernetes, it's solving complex systems. Madhura, great to see you. Thanks for coming back on theCUBE. Thank you for having me again, John. Great to see you. So we're ending up day two here, a lot of action. You guys have something new that nobody has, and we have the scoop here in theCUBE. That's right. So you have something called Elastic Machine Pool, innovative new product you guys built called EMP. I like the acronym EMP, Elastic Machine Pool. What is Elastic Machine Pool? Why are you so excited about it? What, how did it come about? Yeah, no, absolutely. And it's always fantastic to talk about new products here at KubeCon, right? There's just so much energy, a lot of innovation always going on here. We are very excited about Elastic Machine Pools. It's a brand new product off the press, right? And I'm excited to share with you guys here. So what EMP is, is kind of going back to our roots, our core bread and butter has always been about simplifying complexity and providing the best optimization to run your workloads, right? And so we're bringing some of those same innovations that we've been familiar with and we've been part of innovating on the private cloud. We're bringing those to the public cloud now. So specifically what EMP is, it is a brand new compute engine for running Kubernetes in the public cloud. Starting with AWS E-Guess, right? AWS is native Kubernetes service where with EMP you can reduce your costs of running an E-Guess cluster or a form of E-Guess clusters by about 50% or so. What was the problem that you're solving? Where did this come from? Was it because of EKS becoming popular, people are using more of it? Was it not being efficient? Was it underutilized? Was it sprawl? What were some of the problems? Why did this come about and what specifically are you solving for? Yeah, absolutely. So the key word that you use is utilization, right? That's really what we lashed on to and that's the problem that we saw which is pervasive in Kubernetes, especially when running Kubernetes in the public cloud, right? So I don't know if many people don't know this and they get surprised when I talk to them about it which is on average, Kubernetes clusters just across private and public cloud, public cloud especially, tend to be less than 30% utilized, right? So if you think about it, 70% of your capacity across your Kubernetes clusters at any given point is just completely going waste and you paying top dollars, putting that workload, putting that capacity in the public cloud and 70% of those cogs are just pure going waste. So we just had our analyst segment right before you came on and we were talking about the people who are using Kubernetes in the market. 69% of Kubernetes users or companies have Kubernetes running in more than one cloud. So clearly cloud adopts with Kubernetes' key. 56% of those companies have 10 or more clusters. So we're starting to see, as we talked last time you were on the cube, Kubernetes is growing, okay, it's going to become like the Linux, it's going to be invisible and enable other things. What's specifically about the elastic machine pools is relevant. I mean, how do you describe this? What, how would you give the pitch, the elevator pitch to someone, if you say, okay, you got my attention and we do a lot of EKS. I hear people talking all the time, pitching them on about this side and the other thing. What's in it for me, what's the pitch? That's right. And so it's an interesting question because typically when I talk to people about EMP, the next question is, well, I've seen 10 other cost optimization tools in the market, right, how is yours different? And what is very unique about EMP and which is why we're building something that I say nobody else is doing in the market is what we do is instead of you deploying capacity in the form of EC2 instances for your EKS clusters, we go straight to AWS bare metal, okay? That's kind of going back to our roots. We come from virtualization. I spent eight years of VMWare perfecting, helping perfect that. So because we're able to deploy straight at AWS bare metal, now we're able to bring in a ton of innovation, okay? So we put virtualization on top of AWS bare metal. Then we create our own layer of VMs and these become alternate worker nodes. So you go to the bare metal and build from the ground up. The abstractions are tooling or software to manage the pools. That's right. Essentially giving them on demand or like spot instances, I think like spot instances. Yeah, except for, yeah, none of the drawbacks of spot. So instead of you using regular EC2 VMs, you use these elastic VMs, but these VMs are highly, highly optimized with one goal, which is how do we get the best utilization out of that underlying capacity? Okay, and so no one has this? Nobody, not a single vendor in the market has that. This is a unique solution. This is absolutely. All right, so let's get into some of the data that you see from this. What's the result of using EMP or elastic machine pool? What are some of the outcomes that you're seeing share some stats? Okay, first one to do it, I get it, that's cool. What's the results? So the immediate result that most companies you're working with, and we're working with a number of companies right across different sectors, is that we are able to pull up that Kubernetes utilization from 25, 30, 40% all the way up to 60, 70%, right? Typically memory tends to be the most constrained resource and we're able to pull that utilization up. And we're able to do that without any overhead around your application disruption. We don't take your parts down ever because we're using some really cool patent pending technology like live migrating your virtual machines behind the scenes on that pool of Bermuda servers. So the biggest benefit we see with our customers is pulling up that utilization up to about 60, 70%. So your costs just immediately help, right? Like you can do the math and see what a massive impact. Cost drop significantly, what percentage you're seeing cost drop? Roughly. 40, 50%. And utilization goes to 67%. And utilization goes to 67%. So if you take from 30 to 60. Sorry, so no brainer obviously on a work, if it works, I'm assuming it works. So it's access only right now, what's the status of the product? Yeah, so it's early access, but if you go to our website, pluffinline.com, there's a webpage for this product, you can see a video, and the best way to get access is just reach out for a demo. And then we can create a custom report that tells you for your environment, how much we could save. So basically you built a, I won't say a purpose build, but I guess it is, a product specifically to manage Kubernetes clusters in EKS and do it in a way that drives the highest utilization and the best cost scenario. That's right, yeah. Within Amazon, either way, re-invent services. Yeah, well. So what's been the results so far with the customers? Obviously re-invent's coming up. I'm sure you're going to be talking a lot about this. Yeah. Coming up on AWS re-invent. What's some of the feedback you've been getting from people? Give us some taste of a feel for the anecdotal commentary. Yeah, I mean the most interesting comments that I'm seeing as a pattern and across multiple conversations is, I was looking at this product thinking that this is going to be yet another of your cloud health or, you know, cost observability platforms, and I was not going to be that interested, but holy shit, this is very interesting and I would like to know how my team can get an early access. So I'm going to put my skeptical hat on and say, okay, a new service. How does this differ from other compute engines like EC2 and Fargate? Yeah. So EC2 Fargate, you know, they have a completely different approach, right? Fargate is all about getting rid of any worker node, any VM management. So it simplifies operational complexity but Fargate can in fact be more expensive than just running and managing your own EC2 nodes, right? So we provide the same kind of experiences Fargate because we will manage your worker nodes. So with us, you have the same serverless simplicity except at least half the cost of Fargate. So it is comparable to Fargate, but, you know, infinitely more lucrative for an organization. And what's the benefit of the customer versus Fargate? You guys manage it or they have the choice? They have the choice. So they can put some workloads on Fargate, some on EMP. The beauty with EMP is it integrates very seamlessly with existing EKS clusters. So the adoption process is very seamless. So the more clusters you use, if you're going to do a lot of clusters, basically you're saying you believe EMP is better because it's basically built for that. Fargate is just an interface. EMP is better and we've proven that it's better because it focuses on that optimization of utilization which is never a focus for Fargate. Fargate doesn't do that, yeah. What are they optimized for? What's their benefit? Their benefit is how do we delegate away any VM or node management away from you guys, your DevOps teams, for their customers to them, right? So when you use Fargate, you don't have to deal with worker nodes. It's a black box. And there's a simplification in that, right? And we offer that same benefit with EMP as well. So congratulations. Thanks for coming on theCUBE and sharing this scoop on the EMPs. The elastic management will elect the name, I love the EMP, love the acronym. More jargon to feed our AI system. We'd love to promote. I got to ask you, as the Chief Product Officer, since the last time we talked, and you guys have been on theCUBE so many times, going back to OpenStack now, been involved in Kubernetes from day one. I was just arguing with Kelsey, the Hightower, not arguing, but talking about when was the first KubeCon, the CUBE was there and technically it was in San Francisco and then London, so then the CNC, I took it over. I mean, it's been around for a while. You guys have been part of the community for a long time. It's kind of evolving now to, not just about Kubernetes. It's a complex system. Enterprises have to build large-scale complex systems. Kind of the vision that you and I talked about last year of what Kubernetes will talk to. What's surprising you or what observations can you share about this year's KubeCon that's different? Is it, do you agree that this has now gone next level? How has it changed what you guys do and what you guys bring to customers? Yeah, I mean, I think it's been a great continuation and it's gone to the next level because, especially because the focus on AI ML and everybody's expecting the workloads, the resources needed to run them are going to go 10x. So all of a sudden, scale goes to the next order, right? So for example, in some of the conversations, multi-cluster suddenly starts becoming way more important and having some orchestration system that can submit jobs across different clusters. This is a problem that's not been solved in Kubernetes. That's a platform engineering problem. Yeah, exactly. So having proper purpose-built systems to do that or how do we get the best utilization of power because power is going to become one of the most constrained resources. There's not going to be enough power capacity to host data centers, right? That was talked about in the keynote and that's one of the primary topics. So it's definitely going to the next order and I think this wave of AI ML is going to push it even further. Well, it gets our attention. Elastic machine pools, first of its kind solution, targeting a growing area, Kubernetes clusters on the biggest cloud AWS and other clouds probably are on the corner and you can reduce costs and increase utilization. I mean, that's a win-win, all the things line up. You've got to be pretty excited about this product. Very excited, yes, and I urge everybody to give it a try. It's going to be extremely easy to try on. Just go on to our website. If you're watching this, go check it out. Let me know if it works the way she says. I'm excited for the proposition. I can see more growing. Midora, thanks for coming on theCUBE Day 2. Wrap up, I appreciate you spending the time to swing by. Thank you for having me. Okay, that's Day 2 here. I'm theCUBE here, John Furrier, your host. Savannah Peterson, Dustin Kirkland, Joe Peterson, all bringing you all the action here on the ground, Day 2 is wrap. Thanks for watching.