theCUBE Interview with AWS & StormForge

>> According to the latest survey from Enterprise Technology Research, container orchestration is the number one category as measured by customer spending momentum. It’s ahead of AIML, it’s ahead of cloud computing, and it’s ahead of robotic process automation. All of which also show highly elevated levels of customer spending velocity. Now, we drill deeper into the survey of more than 1200 CIOs and IT buyers, and we find that a whopping 70% of respondents are spending more on Kubernetes initiatives in 2022 as compared to last year. The rise of Kubernetes came about through a series of improbable events that change the way applications are developed, deployed and managed. Very early on Kubernetes committers chose to focus on simplicity in massive adoption rather than deep enterprise functionality. It’s why initially virtually all activity around Kubernetes focused on stateless applications. 

That has changed. As Kubernetes adoption has gone mainstream, the need for stronger enterprise functionality has become much more pressing. You hear this constantly when you attend the various developer conference, and the talk is all around, let’s say, shift left to improve security and better cluster management, more complete automation capabilities, support for data-driven workloads and very importantly, vastly better application performance in visibility and management. And that last topic is what we’re here to talk about today. Hello, this is Dave Vellante, and welcome to this special CUBE conversation where we invite into our East Coast Studios Matt Provo, who’s the founder and CEO of StormForge and Chandler Hoisington, the general manager of EKS Edge in Hybrid at AWS. Gentlemen, welcome, it’s good to see you. 

>> Thanks. 

>> Thanks for having us. 

>> So Chandler, you have this convergence, you’ve got application performance, you’ve got developer speed and velocity and you’ve got cloud economics all coming together. What’s driving that convergence and why is it important for customers? 

>> Yeah, yeah, great question. I think it’s important to kind of understand how we got here in the first place. I think Kubernetes solves a lot of problems for users, but the complexity of Kubernetes of just standing up a cluster to begin with is not always simple. And that’s where services like EKS comes in and where Amazon tried to solve that problem for users saying, “Hey the control plane, it’s made up of 10, 15 different components, standing all these up, patching them, you know, handling the CBEs for it et cetera, et cetera, is a very complicated process, let me help you do that.” And where EKS has been so successful and with EKS Anywhere which we launched last year, that’s what we’re helping customers do, a very similar thing in their own data centers. So we’re kind of solving this problem of bringing the cluster online and helping customers launch their first application on it. But then what do you do once your application’s there? 

That’s the question. And so now you launched your application and does it have enough resources? Did you tune the right CPU? Did you tune the right amount of memory for it? All those questions need to be answered and that’s where working with folks like StormForge come in. 

>> Well, it’s interesting Matt because you’re all about optimization and trying to maximize the efficiency which might mean people’s lower their AWS bill, but that’s okay with Amazon, right? You guys have shown the cheaper it is, the more they buy, well. 

>> Yeah. And it’s all about loyalty and developer experience. And so when you can help create or add to the developer experience itself, over time that loyalty’s there. And so when we can come alongside EKS and services from Amazon, well, number one StormForge is built on Amazon, on AWS, and so it’s a nice fit, but when we don’t have to require developers to choose between things like cost and performance, but they can focus on, you know, innovation and connecting the applications that they’re managing on Kubernetes as they operationalize them to the actual business objectives that they have, it’s a pretty powerful combination. 

>> So your entry into the market was in pre-production. 

>> Yeah. 

>> You can kind of simulate what performance is going to look like and now you’ve announced optimized live. 

>> Yep. 

>> So that should allow you to turn the crank a little bit more. 

>> Yeah. 

>> Get a little bit more accurate and respond more quickly. 

>> Yeah. So we’re the only ones that give you both views. And so we want to, you know, we want to provide a view in what we call kind of our experimentation side of our platform, which is pre-production, as well as on ongoing and continuous view which we kind of call our observation, the observation part of our solution, which is in production. And so for us, it’s about providing that view, it’s also about taking an increased number of data inputs into the platform itself so that our machine learning can learn from that and ultimately be able to automate the right kinds of tasks alongside the developers to meet their objectives. 

>> So, Chandler, in my intro I was talking about the spending velocity and how Kubernetes was at the top. But when we had other survey questions that ETR did, and this is post pandemic, it was interesting. We asked what’s the most important initiative? And the two top ones were security, no surprise, and it popped up really after the pandemic hit in the lockdown even more prominent and cloud migration, 

>> Right. 

>> was number two. And so how are you working with StormForge to effect cloud migrations? Talk about that relationship. 

>> Yeah. I think it’s, you know, different enterprises to have different strategies on how they’re going to get their workloads to the cloud. Some of ’em want to have modernize in place in their data centers and then take those modernized applications and move them to the cloud, and that’s where something like I mentioned earlier, EKS Anywhere comes into play really nicely because we can bring a consistent experience, a Kubernetes experience to your data center, you can modernize your applications and then you can bring those to EKS in the cloud. And as you’re moving them back and forth you have a more consistent experience with Kubernetes. And luckily StormForge works on prem as well even in air gapped environments for StormForge. So, you know, that’s, you can get your applications tuned correctly for your data center workloads, and then you’re going to tune them differently when you move them to the cloud and you can get them tuned correctly there but StormForge can run consistently in both environments. 

>> Now, can you add some color as to how you optimize EKS? 

>> Yeah, so I think from a EKS standpoint, when you, again, when the number of parameters that you have to look at for your application inside of EKS and then the associated services that will go alongside that the packages that are coming in from a Kubernetes standpoint itself, and then you start to transition and operationalize where more and more of these are in production, they’re, you know, connected to the business, we provide the ability to go beyond what developers typically do which is sort of take the, either the out of the box defaults or recommendations that ship with the services that they put into their application or the any human’s ability to kind of keep up with a couple parameters at a time. 

You know, with two parameters for the typical Kubernetes application, you might have about a 100 different possible combinations that you could choose from. And sometimes humans can keep up with that, at least statically. And so for us, we want to blow that wide open. We want developers to be able to take advantage of the entire footprint or environment itself. And, you know, by using machine learning to help augment what the developers themselves are doing, not replacing them, augmenting them and having them be a part of that process. Now this whole new world of optimization opens up to them, which is pretty fantastic. And so how the actual workloads are configured, you know, on an ongoing basis and predictively based on upcoming business events, or even unknowns many times is a pretty powerful position to be in. 

>> I mean, you said not to replace development. I mentioned robotic process automation in my intro, and of course in the early days, I was like, oh, it’s going to replace my job. What’s actually happened is it’s replacing all the mundane tasks. 

>> Yeah. 

>> So you can actually do your job. 

>> Yeah. 

>> Right? We’re all working 24/7, 365 these days, so that the extent that you can automate the things that I hate doing, 

>> Yeah. 

>> That’s a huge win. So Chandler, how do people get started? You mentioned EKS Anywhere, are they starting on prem and then kind of moving into the cloud? If I’m a customer and I’m interested and I’m sort of at the beginning, where do I start? 

>> Yeah. Yeah. I mean, it really depends on your workload. Any workload that can run in the cloud should run in the cloud. I’m not just saying that because I work at Amazon but I truly think that that is the case. And I think customers think that as well. More and more customers are trying to move workloads to the cloud for that elasticity and all the benefits of using these huge platforms and, you know, hundreds of services that you have advantage of in the cloud but some workloads just can’t move to the cloud yet. You have workloads that have latency requirements like some gaming workloads, for example, where we don’t have regions close enough to the consumers yet. So, you know, you want to put workloads in Turkey to service Egypt customers or something like this. You also have workloads that are, you know, on cruise ships and they lose connectivity in the middle of the Atlantic, or maybe you have highly secure workloads in air gapped environments or something like this. 

So there’s still a lot of use cases that keep workloads on prem and sometimes customers just have existing investments in hardware that they don’t want to eat yet, right? And they want to slowly phase those out as they move to the cloud. And again, that’s where EKS Anywhere really plays well for the workloads that you want to keep on prem, but then as you move to the cloud you can take advantage of obviously EKS. 

>> I’ll put you in the spot. 

>> Sure. 

>> And don’t hate me for doing this, but so Andy Jassy, Adam Selipsky, I’ve certainly heard Maylan Thompson Bukavek talk about this, and in fullness of time, all workloads will be in the cloud. 

>> Yeah. 

>> And I’ve said the cloud is expanding. We’re going to bring the cloud to the edge. Edge is in your title. 

>> Yeah. 

>> Is that a correct interpretation and obvious it relates to Kubernetes. 

>> Absolutely. 

>> And you’ll see that in Amazon strategy. I mean, without posts and wavelengths and local zones, like we’re, at the end of the day, Amazon tries to satisfy customers. And if customers are saying, “Hey, I need workloads in San, I want to run a workload in San Francisco. And it’s really important to me that it’s close to those users, the end users that are in that area,” we’re going to help them do that at Amazon. And there’s a variety of options now to do that. EKS Anywhere is actually only one piece of that kind of whole strategy. 

>> Yeah. I mean, here you have your best people working on the speed of light problem, but until that’s solved, sure, sure. 

>> That’s right. 

>> We’ll give you the last word. 

>> How do you know about that? 

>> Yeah. Yeah. (all laughing) 

>> It’s a top secret. Sorry. You heard it on theCUBE first. Matt, we’ll give you the last word, bring us home. 

>> I, so I couldn’t agree more. The, you know, the cloud is where workloads are going. Whether what I love is the ability to look at, you know, for the same enterprises, a lot of the ones we work with, want a, they want a public and a private view, public cloud, private cloud view. And they want that flexibility to, depending on the nature of the applications to be able to shift between from time to time where, you know, really decide. And I love EKS Anywhere. I think it’s a fantastic addition to the, you know, to the ecosystem. And, you know, I think for us, we’re about staying focused on the set of problems that we solve. No developer that I’ve ever met and probably neither of you have met, gets super excited about getting out of bed to manually tune their applications. And so what we find is that, you know, the time spent doing that, literally just is, there’s like a one-to-one correlation. It means they’re not innovating and they’re not doing what they love to be doing. And so when we can come alongside that and automate away the manual task to your point, I think there are a lot of parallels to RPA in that case, it becomes actually a pretty empowering process for our users, so that they feel like they’re, again, meeting the business objectives that they have, they get to innovate and yet, you know, they’re exploring this whole new world around not having to choose between something like cost and performance for their applications. 

>> Well, and we’re entering an entire new era of scale. 

>> Yeah. 

>> We’ve never seen before and human just are not going to be able to keep up with that. 

>> Yep. 

>> And that affect quality and speed and everything else. Guys, hey, thanks so much for coming in a great conversation. And thank you for watching this CUBE conversation. This is Dave Vellante, and we’ll see you next time. (upbeat music)