 Hey, everybody. So I am not Evan Brown. Does anybody know Evan Brown by chance? Couple people. OK. So Evan is the one that's actually doing most of the work on this, so he couldn't make it, but I'm filling in for him. But I wanted to put his name up there. And I guess a little bit about me. I run a small team at Google Cloud working on open source integrations. And so recently, we've gotten involved with Pivotal and Cloud Foundry and Bosch. So I'll talk a little bit about Bosch CPI work, what we've done so far. So a little history about the Bosch CPI. How many people have heard of Ferdy? Woohoo, lovely. That was a mistake. How many people have not heard of Ferdy? Yeah, OK. There's two or three maybe. So Ferdy actually, so I only went back about a year, but actually the history of this goes back even further than that. So the very first time that Ferdy fiddled around with creating a Bosch CPI was back in Ruby. Last year, he decided, hey, I want to learn how to play with Go. So he had a little small 20% project to develop the CPI for Google written in Go. And then you can see it kind of sat there for a few months or more than that. And then that's roughly when Google kind of started getting engaged. And we had one of our solutions architects kind of starting to work with Ferdy just to brainstorm a few things that we might make some improvements on. And then it wasn't until really about a month and a half, two months ago, where Evan and I really started to get involved. We came down, we met with Pivotal, talked a lot about what we could do with the CPI. So that's kind of where we're at today. There's a link to it. You can go check it out. We've got instructions for how to go ahead and get it started up with Bosch in it and everything. So I'll talk a little bit about the features that are in there. We've got, I don't see the slides down here. So some of this stuff would sound a little different because you maybe not are familiar with some of the Compute Engine features. One of the things that we did was we removed the Bosch registry and we replaced that with GCE metadata service. So what's different about, oh, thanks, on behalf of Evan. So GCE has writable metadata. So it's a good way to sort of pass information back around between the virtual machines that are in the environment and the director. So that's one of the changes that we made. And I think that that was suggested by either Ferdy or Dimitri or somebody like that, that that would be a good way to make this a little bit more scalable and resilient and leverage some of the infrastructure innovations that we've made with GCE. Manual networks has just recently gone in and being able to specify private IP. So this was a new feature that just rolled out not too long ago. We've got subnetworks in GCE as well. So all of those things have now made it back into the CPI and it's actually merged in. Work in progress. So this is kind of one of the things that Evan has been spending a lot of time on recently is how do we use GCE's load balancer and make it kind of look like it behaves like on other cloud providers. So we've gone through and we've used our Layer 7 load balancer, which does URL routing. But that stuff's not really exposed yet, but it all gets plumbed back through into Bosch so that when instances come up, they join the load balancer automatically. Disc attachment improvements, things like that that we're still working on. To do, I went through Bosch training and realized like Bosch SSH doesn't work. So there's a lot of little probably pain points that we're still finding as we go through this. Evan and I will actually be here next week working with Pivotal to kind of identify some more of these potholes and come up with a punch list of things to work on. Blobstore on GCS probably just works out of the box. We have an S3 compatibility mode on GCS. So that should be an easy one. Pivotal's building a prototype GCE stem cells right now. So all of that's going really well. We've got the concourse pipeline set up. We build and release the CPI. We also do a light stem cell. We pull the heavy stem cell across from S3 that Pivotal builds. And we drop that into GCS. So as you spin this stuff up on GCE, you've already got that stuff inside of Google's network. We went through added a whole suite of live integration tests on that. So it takes about an hour to run. Costs you like $20 or something. So maybe you don't want to do that as a starting point. And then we're also passing bats. So I want to tell you a little bit about Google Compute Engine. So I'll keep it kind of short. There's a couple of cute things. Nice things I'd say that I want to call to your attention. So one of the things is we have discrete machine sizes. But we also have customizable machine sizes. So if what we have as static representations in terms of CPU and memory doesn't fit your needs, you can actually customize that and create your own. Advanced networking. So I was telling you a little bit about that. The load balancer that you would spin up in front of your Cloud Foundry instance is the same load balancer that we use internally for Gmail, YouTube, Maps, all of that stuff. So there's a lot of power there in terms of our networking. And then also talk about how much cheaper it is. It's sustained use pricing. So the longer these instances run, the cheaper they get. And lastly, this is where you can find B, if you have questions. And then also, I encourage you to get started with a free trial. And then there's also the link to the repo again. So that's it. I'm out of time. Next up is Tushar.