 Welcome. Today, Jason and I will discuss the multi-cloud maturity model, the articles linked in the description. And Jason had a few questions, so I'd say fire away, Jason. Yeah, so it was great articles, thanks for sharing. The thing that really struck me, though, is I was reading this book. First of all, it sounded like a great maturity model that really made a ton of sense when I was thinking through these problems myself. But in another company that I worked at, one of the things that we were thinking about was when do you make this decision in order to go from get into the cloud, go quickly as possible, take advantage of all of the great stuff that AWS or GCP or anybody else offers in order to get you up and running quickly, versus starting to plan for the eventuality that you're going to want to be in a multi-cloud environment. We really struggled there because we wanted to get to market fast, so we wanted to take advantage of things like managed services and things like that that wouldn't be available to another provider. And it wasn't really clear to us when was the right moment to start planning for the other. We sort of felt like we got too far down the path of taking advantage, where it was making it more difficult for ourselves later. But then again, we got to market quicker. I'm curious what your thoughts are on that. Yeah, that's great. I think that decision is getting a lot easier. And that's thanks to open source. So it used to be like a couple of years ago, like you had no other option if you wanted to manage service, you had to get something proprietary. And the examples of that are Kinesis, the AWS service that makes it easy to analyze streaming real-time data. And another example would be DynamoDB, an Amazon service that makes it easy to have databases that have massive scale and that have a lot of read and write traffic. And I think that is changing. And that is changing due to open source. Just to reinvent, Amazon launched Managed Kafka. Kafka is a direct alternative to Kinesis. And so Amazon is switching from proprietary alternatives to open source alternatives. And the same thing I think is happening in the database market where DynamoDB recently introduced assets transactions with a whole lot of conveyance while players like Citus Data already offer you the ability to scale Postgres massively. So you could just go with Kafka. You can just go with Postgres and then scale it up later with Citus when it's needed. So I think that makes the decision to go for application portability to have your applications being able to be deployed on any cloud a lot easier since there's now managed alternatives based on open source that will allow you to go to any cloud. That's very cool. Yeah, it wasn't the case a couple of years ago, for sure. What we were looking forward to the most was the Amazon Kubernetes offering where we could consider Kubernetes was sort of our interface to the cloud where we could use GCP and AWS in a consistent way. But I'd heard that those solutions were sort of diverging to some extent. I wonder if you've heard any update on that. Yeah, I think the CNCF has introduced Kubernetes certification. And I think all of the major players are certified. I do think that there's differences. For example, EKS doesn't set some defaults. That costs, for example, for GitLab, some of the deployments to fail if you haven't said those deploys. Those deploys are set on most other Kubernetes offerings. I think over time, these offerings will go towards each other. I think right now, the one that's most out there is the one with the most history. It's an open shift. Open shift got started way before Kubernetes. They have a lot of things that are an open shift. But over time, we're added to Kubernetes that they still have to switch out. I know a couple of months ago, Helm charts were still a permission problem. All not due to any Eileen Malus or anything like that, but just due to Red Hat being great at security and having thought about this and having made this before Helm became a standard. But I think everyone is going to demand vanilla Kubernetes. And I think Haptio, for example, has been really successful with that. They just got acquired for hundreds of millions of dollars. And I think that's what people want, manage vanilla Kubernetes and that's happening. Now, I don't think Kubernetes solves right now all the other things. You still think I want to manage servers for everything that's stateful. And I'm really excited today with the launch of Crossplane, an initiative by Basamp who's making Rook as well. But he's trying to make sure that not only can you manage Kubernetes with custom builds with CRDs, but also all the stuff around it so that you can also request a managed data store or a DNS record or an IP address in a multi-cloud way. And I think that's, Crossplane is a really exciting development there. Interesting. Where do you think the, so if these services become more consistent and more commodified across the providers in order to give you this ability to be multi-cloud? Where do you think the providers are going to differentiate with each other? Yeah, I do think that there always will be a frontier. Right now, it's AI, ML, and Amazon just gave me a deep lens thing to experiment with. They're all going to pour a lot of resources in making those things better. So I think, like usual, the proprietary things come first, and then open source comes after that to kind of make it more accessible to everyone. I think that this will be no different than there's always a frontier. Amazon is way ahead of everyone else with Lambda. What they're now doing with Lambda layers is just awesome. And well, the open source is following, but they're clearly ahead. And there will always be frontiers to explore. Also, I think the competition on network latency, on price, on reliability, on how good can you sell? What kind of creative constructs can you do in a deal? We'll get more important. There's computers at big cost to companies if they can shift that into the future or pay it in a different way. And that's going to be really, really interesting. And I expect huge discounts if you're able to say, OK, if you give me a better offer, I'm going to leave the other cloud for 99%. I think that that will put you in a much better position as a company, because this competition is going to heat up. Oracle had cars out doing reinvent that says, we cut your bill in half. That is that the big three are not engaging with each other like that yet, but that is going to come. Yeah, cool. And then in terms of the administration and, I guess, control planes is the right word for it, of how companies are managing across multi-cloud providers. Do you see anything interesting happening in that space? Yeah, there were a couple of tweets, I think it was from Kelsey Hightower, that, yeah, it's not one control plane. It's like multiple ones in a nested control plane. It's getting a bit hard to manage that. I love what Google's doing with GKE on-prem, where they say, look, we're not just supporting GCP, but also on-prem. I do think the future is not just hybrid, but multi-cloud, where it's not just on-prem and one-cloud, but many clouds, including your on-premises cloud. I'm, again, very excited about cross-plane there. That is a way to kind of manage everything with Kube-Cuddle, but where you can also provision Kubernetes with that and all the managed services and everything else. But this is going to be a very interesting space, but for now, a lot of creativity, some say chaos, and we're going to see a winner emerge in the next couple of years, or maybe multiple ones. I think Red Hat has been out in front of everyone with OpenShift offering. Pivotal is doing a nice job now transitioning to Kubernetes, but at Reinvent, for example, NetApp even had a Kubernetes distribution. Cisco had one. Everyone has the Kubernetes distribution. The next thing will be to have a multi-cloud control plane. So that's going to be interesting to watch. Yeah, for sure. Well, I wonder if I can ask you a bonus question as to where AWS Reinvent and I was following along remotely. Is that OK? Yes, for sure. A lot more directly related, but I'm curious what your thoughts were on Firecracker and what that's going to mean. Yeah, so Firecracker is a new AWS open source project. It's actually based on a Chromium project. It's a direct fork, so I think there missing attribution there, but it's a way to launch containers faster in a more secure way. And what we're seeing here is that Amazon is getting a handle on a cold start problem of Lambda, and they're doing an awesome job there. So Lambda used to suffer from a major cold start problem where anytime you scaled or anytime something was replaced, like at unpredictable intervals, a request would take a lot longer because you still had to create a new container and load your zip files and everything else. And they're laser focused on it, and they're doing these kinds of things to get rid of it. And I think serverless is going to go from more of a feature or a specific thing to more of a spectrum. And on one hand, you have an application that's running all the time. On the other hand, you have something that's completely serverless that doesn't have any processes running when you're not using it and anything in between. And you can tune yourself or it's automatically tuned for you where you want to be and how much cold start you can accept. And you're already seeing that with the elastic load balancers where you can say, hey, I want 40% or 50% or 70% utilization of my service. And that's based on, are you human facing? Is there a human bothered by the cold start? Or is it a background process that doesn't care? So we're going to see more of that. Very cool. That's a good explanation. I was following along in Twitter. I could pick up the excitement. But I couldn't get quite a simple explanation like that. So that was great to hear. I think that was all the questions I had said. Cool. Well, Jason, thank you very much. Yeah, very nice chatting with you. Right. Talk to you.