Moving up the stack, talking Pivotal Cloud Foundry 1.6 James Watters

Episode 4 · 2015-10-29 · 41:59 · SRT · archive.org · all Tanzu Talk

Well, hey, James. How's it going? Hey, Mr. Coté. It's a pretty excellent morning here in New York City. I heard that you had an experience of using a regular cab. How did that go? I mean, Uber is a trope, if you will, of representing our company. So almost metaphorically, you got to go experience what the old school way of doing stuff. I think it's a good analogy. How did it pan out? How's the technology in there? I hit a wall last night. I take great pride that I travel a lot, but I never complained about traveling on Twitter because it's sort of like this default that my whole Twitter stream is full of extremely well-paid people complaining about small travel inconveniences. I can't believe an important person like me is forced to wait for peanuts. You know what I mean? Like these kind of tweets. Dear God, this scotch is only 18 years old. I don't know how I'll cope. Exactly. So I'm like, I've always tried really hard. I'm like, that is like the death of interesting information to share. But after a long delay, I got into JFK at 3am and I am addicted to Uber. I fully admitted I Uber everywhere. I really never used my car. I am a San Francisco person. And I tried to connect to restarted my phone. I tried to connect. I think their authentication service was just down and it was down for about 20 minutes last night in New York. And so I crawled into the back of one of these yellow things. And as you described, it's sort of like a hamster cage with a plastic wall. Exactly. But there's no tubes. It's just the square cage, the starter set. Yeah. Yeah. And it was funny because, you know, my whole quest is like, hey, telling the world about reliable next generation cloud software. And when we say software is eating the world, Uber is out to get you. And last night, Uber had some issues scaling their software. So no one's perfect. But I've actually heard that they're refreshing their tech team there. And it's often the case,

sometimes you get one hit app and you're not amazing at scaling it forever. You know, remember when Uber came out? I mean, sorry, when Twitter came out, they had a killer app, but not a killer tech back end yet. Yes, those were fun times. The whale. That's right. That's right. Yeah. You know, I spend a lot of time when I'm traipsing about with the various, I don't know, one would call them enterprise teams and groups that I talk with. And I have this conflicting notion in my head that there's something valuable in the idea of fail fast, but I don't think you necessarily get permission to do that for years on end. Like there's a lot of tolerance that you're given at first to learn how to do things. And then there's an expectation of just executing seamlessly. But yeah, that's a very paradoxical notion. That's almost like it almost has a Zen quality of how you react to it ongoing as more of a Rorschach test about what you should be doing. If the idea of failure equals learning, if that resonates with you or sounds scary, but it's good calisthenics to put your mind through. Yeah. I would say in the cloud, you want to fail fast and then heal exceptionally fast. That's right. Fail faster and fix faster than that. Like if I had, you know, you know, after having restarted my phone and tried to connect for the third time gotten through, I think, I think I totally forgive the situation. Right. I'm like, oh, they had a momentary blip. Things self healed. Exactly. That's yeah. We, we always, I always try to say instead of, uh, uptime, I try to focus on resilience. You want to have that, uh, you want to be able to bounce back quickly. That's right. So we were getting together today. We, we have a, uh, we have a new release of Pivotal Cloud Foundry or as we might slip into, uh, PCF, uh, coming out version 1.6. And, um, you know, we have, we have a flurry of details out there, uh, about what's in it. And there's, there's a, there's a bevy of features. I was just watching an internal webinar we have and, uh, it's chock full, but what I want to talk to you about is, you know, I've been here since, uh, January of this year and, and we recorded a podcast back in, in May, I think, or when we had

1.4 out. That's right. And I feel like in, in so much has evolved since then and, and a huge part of, of what's evolved is, is what P well, I feel like what, what people, what our customers are looking for, right? Like, like, uh, now, you know, I, I don't go out as much as you do and get trapped in a hamster cages as it were. But like when, when, when I go out and talk with people, even over that short amount of time, the problems that they're looking to solve have very rapidly moved up the stack. And, and what I mean by that is back when I was at a four or five one and, and even doing, uh, M and a stuff, uh, at Dell, when I was there before that much of the focus on cloud was basically on like, just getting your infrastructure up and running and almost the, uh, almost the cool of that and getting all your stuff doing going. And, and you still hear a lot of conversation about that, like, uh, you know, container scheduling and orchestration and sorting out what that is. And to a limited extent, like people that I talk with nowadays at large organizations still go over that, but like very rapidly, once you, they're kind of assured that you have the infrastructure, they're actually more interested in just like, I don't know what you might call good old fashion application development. That's right. Exactly. I find my, that's, that's it. That's a good, that's a good poll. Cause I find myself saying ALM, like you remember ALM or, and, and, and then the second level of seeing if they remember app dev stuff is like, do you know what SDLC is? You remember that? And anyways, so again, like even, even when I, uh, uh, joined Pivotal back then, that conversation was still very, very op centric and infrastructure centric and looking, looking over all the, what I would call the major features that we have in, uh, Pivotal Cloud Foundry 1.6. It's interesting that things like, uh, spring boot and spring cloud, uh, you know, depending on what product manager and things like that, they arguably kind of dominate the interestingness of, of, of what's in there because

they're enabling doing, doing cloud native application development. So that that's, that's my, uh, my monologue to a question. And, and, and, you know, I, I think, I think what you're in an interesting position being basically the, the GM of all of this, the general manager of all this to go over is like, I wanted you to kind of walk us through the story of how we got in the short amount of time from basically the ops layer of stuff all the way up to basically things like boot and cloud being what I think is one of the main centerpieces of, of what we're doing with Pivotal Cloud Foundry. Yeah, it's, uh, it's, it's, uh, it's been a really fun transition. Um, and the thing I'd say is, is that I think we got smarter and customers are getting smarter and the market is slowly getting smarter that the automation problem they were trying to solve, it's best to trace it back to its root cause. You know, so I think what's, what's happened and I'm working on a personal blog about this, you know, help me. I'm, I'm not the writer that you are, Michael, but, uh, I, I challenged myself to share my own thoughts sort of independent of, you know, the company I work for and the products I work on, but like what's really going on out in the market. And at a, at a really high meta layer, when I think about it, at least it's sort of like SAS software came along and tried effectively to take away a lot of the very undifferentiated software and have other people operated for you. Right. So you can go to work day. Now you can go to Salesforce. Um, you can use all kinds of SAS apps concur and you can run a lot of your business. I mean, you, you're at Pivotal now. Like we, we don't do a lot of like app hosting internally, right? Like, can you think of an app that we use that's run by it? No, no. I mean, I mean, we're, uh, as a testament to that, I'm very familiar with the Okta supposed single sign on screen. The Okta, please do Google authenticator again, please, sir.

Yep. But I, you know, that's, that's, that's just symptomatic of the point is that, um, we do have some on-premise stuff. That's our own software of, you know, things that we do on our own, but I'm pretty sure most all of our backend is, uh, all SAS, SASified. That's right. So we, you know, we were like, when we started Pivotal, we're like, Hey, let's, let's pretend like we're starting not pretend let's, let's do a blank slate. We're starting a new company here happened to be a spin out of some big existing companies, but we really went greenfield with it. And so what, what you end up seeing is you see that a lot of what it used to do is just like sort of, you know, I joke when I started at Sun Microsystems back in the dark ages in 2001. Um, that my first day at work, I talk about this is SpringOne, my first day at work, I was all excited. I've been into Java development and really cared about that. And I thought all this is going to change the world. And I got, got a job at the big dog sun. I was so happy. And, uh, I was doing this rotation where they show you different things that happens at sun. It was cool. I was like, okay, first day, what are we going to go do this week? They're like, well, we're going to do with this customer, the server sizing assessment. I said, what? I said, well, we're going to spend the next six weeks assessing how big of a server they should buy for this app they're writing. And there's this very like crawling, plotting 18 month pace to everything that was happening back then. Um, because a lot of it exists in a world where they'd spend 12 months or 18 months standing up people soft or standing up Oracle, you know, e-business or any of these big, you know, suites that are now often bought as SAS. And so a lot of it was really about like, how do you automate or how do you control all of these, you know, off the shelf pieces of software that come your way? Right. Right. And I think when I think back to root causes, like, that's what kind of got, that's the transition run right now is that the blog I want to write is called life after SAS, which is like it's job

now is more and more and more going to be to run the differentiating software that that company wrote. Like if it continues to exist, it's facilitating security for SAS and access, and then helping run the software that the company wrote. And then the question is, is that now that you have an opportunity to, to focus on that and really focus on it, and we've learned everything we've learned since the early 2000s, are there patterns of software development that enable, you know, hyper-efficient, you know, continuous delivery and scale? And I think the answer is yes. Right. And I think that's why we've really started to focus on, okay, now your job is to run custom software. How do you do that? And how do you think about that holistically? Exactly. I mean, I mean, I, I think that's the, uh, that that's, yeah, I, I, I reduce this down to an equation that, that, you know, it's not a real equation, but, but I, I, I write it up as IT minus SAS equals what? And you just basically define the what, right? That once, once you remove SAS from the big world of IT, what's left to do and what's left to do is our, you know, arguably you got to set up the network and maybe there's some desktop management, but that ruins my story. So I ignore that part, but, but essentially it brings up, well, there's writing your own software. And, and I think, I think that is, uh, you know, we, we at Pivotal have always been on about, uh, you should be really good at writing software, right? Cause that's what we sell. And that's what people want to use us for is the ability to do your custom software. But with the framing that, that you are walking us into, it becomes clear that that is what IT is basically important for now, other than just like outsourcing your stuff to SAS. And I think, uh, yeah, I mean that, that kind of framing tends to, uh, resonate pretty well with people. And then the thing that I find, I mean, tell, tell me if you, uh, you find this as well, but once you walk someone through to that, you almost have to have them unlearn their IT behavior and practices of the past. And the way

that I tend to frame this is like for a long time, IT's job was about optimizing how all that on-premise IT was run. Right. And, uh, we, we being part of, you know, the EMC Federation directly benefited from that with VMware, right? Like VMware conquered that and, and essentially optimized how a huge part of how IT was run. And now that IT there's ways of doing, of optimizing how IT is run. And then you also have SAS, the next sort of 10 years or whatever is figuring is returning back to writing your own software. And, uh, I think, uh, you know, not all companies are like this, but a lot of large organizations that I ended up talking with have kind of forgotten that craft and, uh, they need, they need to return back to how to start doing software. And then as you know, a lot of what we, we provide is they need the, the frameworks and, uh, the actual tools that help them do it. Yeah. And I, and I think that there were pockets of people in those companies that knew how to do it, but what happens is that when you treated every app deployment and every off the shelf app management, like its own project, right? So it all got its own build out. And this is where the world of trouble tickets and everything came from, right? Pre-cloud essentially. Everything was a one-off and therefore every custom developed app was treated as a one-off and got its own environment. However, it wanted, I mean, I spent Monday with a really talented team of, um, you know, banking, uh, CIOs from one of the top 10 banks. And, uh, you know, their challenge was that every, they're trying to build a cloud really to run all these things at massive scale. And they have a ton of custom software, like massive amounts of it. But in the past, because of the mix between running off the shelf and running custom, there was sort of not attention to unifying custom, um, and the way a company like Facebook or Google or Netflix would pay hyper attention to the infrastructure for every custom app. Um, what happened was that every business came at that, came at them with a different new way of doing

everything. And so, you know, not to plug our stuff too much, but when they found out about spring boot, which is a way of, you know, writing, you know, next generation Java applications without a lot of variation in terms of how the application and, you know, server environment are set up, um, very homogenous approach to it, you know, they were thrilled. And that's where a lot of our businesses come now just from getting to the app dev teams. I mean, there were some beautiful tweets from Comcast and they were public about it, about, we started teaching them spring boot and spring cloud. They're like, wow, this is really exciting. This is a better way to write custom software. And then the, the hidden surprise and the hidden treat and what makes Pivotal such a special company now is that then there's a platform that can operate that at super efficiency, high availability. So it really is, I do think a new way of looking at things, um, you know, holistically from the app team, the whole way down to the, to the OS. And I think, I think, you know, there's, there's another way I've been metaphorizing in my head is that the, there's, there's been a lot of effort and software and innovation and work as it were put into automating and perfecting how you operate infrastructure, right? Like I, I, I, I, I'm, I'm on this podcast, uh, some cloud podcast later today, and I know they're going to ask me what the cloud is. And I've been thinking about what answer I'm going to give them. Like I should probably have a pat answer I give to everyone at this point, but I don't know what I'd say. That's a good question at this point. Exactly. It's, it's, it's a question of like, Oh my, I've, I've been studying this for six years and every day I get more confused, like in, in a good way, but it's sort of like, anyways, it's, it's a sort of a, an expert problem, I guess. It's like asking what poetry is or something. Um, but I, I think, I think a large part of it is, um, is, is at the infrastructure layer. Again, you basically standardize and automate as much as possible so that you're not manually setting up and managing your infrastructure. Like you always need to manually do some things, but the, the big thing

that, that cloud at the infrastructure layers tends to do is just automate lots of stuff and remove variability. And then the benefit you get from there is it's easier to manage. And if it's easier to manage, you can do things with it faster. So when, when, when I've been, when I've been looking at the rise and importance of spring boot and spring cloud, I I've been thinking and, and, and you launched me on this tangent, uh, with your explanation of people reacting to it is we're almost bringing that notion to the application development middleware layer, where instead of developers coming up with every time they do a new project or a product, a new way of operationalizing their middleware services, the fully integrated stack that we have and how spring boot and spring cloud fit into it essentially standardizes that. And then you don't have to worry about how you hook together all these microservices and manage them and how you, how you use something like spring boot to package up the services that you have. And, and, and I think in this it minus SAS equals what sort of world, it's almost like that starts to be the job that operation needs to do is how do I run this middleware, which I don't think is really a conversation that's happened in it land in a long time. Like, how do I run that software? It's always been customized and one off. And, uh, I don't know. I mean, I, I feel like a lot of what we're doing in one six is, is getting towards that. Like here's a standard set of middle. Well, it's not even a standard set of middleware. Here's a standard way to run middleware that allows you to not worry about how you operationalize it and run it. And in fact, I think the great thing, and I think where people had tried to come at this before was they, they picked seven templates of middleware. So like, here's seven formatted VMs. We call this IaaS plus that you're allowed to choose from, but then they just handed those VMs with that middleware on it to the teams. And that's not really middleware as a service, right? You're like, well, how would I bind service discovery to one of these nodes so that it can find the other ones? Well, that's left to the developer to think about. Yeah. Those are just, uh, to use our metaphor, those are just hamster cages

you're putting people in. Like it doesn't really let them create outside of the hamster cage. I, I think having gone through the pain of looking at that glass, yes, that's not fun to look at that glass. It's, it's sort of like, Hey, this is a little limiting. Um, and that's where it kind of got a bad reputation of like no one wanted what it gave them. Um, and I think people have been really excited because we've become something that developers get excited about getting from it and put it in a simple sense, simple sentence. Yeah, no, I think that's, that's, uh, that's definitely the case. So, so let's, you know, there's another aspect. Let's, let's go back to the, we were, we were joking about, uh, ALM and SDLC. And that's another, uh, another clutch of things we do. What we we've been doing now in, in the reference architect, I mean, in the cloud area, I would almost call the type of work we've been doing. They're almost a reference architecture. That's sort of integrated. We have, we have an opinionated tool chain that you could put in place in the end of what you would need for your, your sort of whole ALM process essentially. Yeah. So like, let's get explicit in one, in the one six release, uh, in November, we've integrated three partners, get lab, artifactory, and cloud bees Jenkins. Um, and we're also bringing Pivotal tracker by the way, um, which is not an external, but it's part of the ALM process that Pivotal labs is so good at, um, directly to the platform in a turnkey way. And we've done the same thing, which is we've, we've brought cloud automation, scaling, unified identity management, et cetera, to all of these things such that you just push a button on the platform. And now you've got, you know, scalable, high availability version control system integrated into your platform with GitLab right there. Um, and so we're taking it even to the more extreme end. And I think people, people get that they just want to write code and then hand it to a system.

Like, I think everyone gets that. And in a sense, Cloud Foundry has always been about that as like, here's my code running the cloud for me. I don't care how it was the beautiful little haiku that Unsee wrote about it. Um, but you know, there's something that happens before you've got that code already packaged and ready to deploy. And those things look like version control, uh, artifact and build management and, uh, CI systems. And so that's what we've, we've had to add to the system. So now you can kind of come from your very first moment of composition, um, and GitLab ends and on the system. I think that's, that's, again, that's really focusing on like, look, our job is to deliver custom software. So let's not leave a rock unturned and helping you do that. Yeah. And I think, I think it's also a, uh, I mean, it'll be fun to see how we evolve it over the next, the next year, essentially. And more importantly, how people use it and adapt it and how it pans out. But it is, uh, we, we, we talk a lot about, you should treat your continuous delivery pipeline as a factory. Like that's, that's your primary asset essentially that allows you to produce things. And it's, it'll be fun to see what having a unified end to end thing looks like. And, and, and, and, and essentially like how, how that evolves because up until now, like, I mean, I mean, arguably if you were a, let's say Microsoft developer, historically, you sort of had your MSD and CDs and you just had everything, like everything from one source, all integrated, it all fits together. And that's why, even though they were closed source and expensive, they did well for a while. Exactly. And, and, and it'll be fun to see if we can bring that, that beneficial best practice into the general world, right? Like, and, and if, if we can have a fully integrated pipe end to end, cause man, you could spend a lot of time integrating that together. And, and, uh, and I mean, basically waste a lot of time on all the differences between all the installs of it. I talked to one customer, they're like, we have five version control systems, five build systems,

and all of it runs through custom trouble ticket deployments with IT in the middle. And we have some code running in production. And this is the blind that terrified me. And we're not sure how it got there. Like, we don't know what maze through these systems it took to get there. Yeah. Yeah. No, my, my, uh, my, my, my cherished anecdote along that line is I remember I was talking with someone at, at, uh, at a, at a large, a large company and I had to slow them down the conversation and like, let me just, let me just explain back to you what I think you just said. You basically wrote, uh, a chef script to open up a ticket for this part of your build pipeline, right? Like, like in order to automate the whole pipeline, there's still a help desk involved and you fully automated that. And indeed that's what they had done, but, but you know, it's, it's, it's, it's that motley sort of chain. And, and with all of this stuff, once you optimize the full end to end process, I think, I think there's interesting new benefits that, uh, only the sort of unicorns or whatever have appreciated that that will be bringing to the, the, the rest of the world. Yeah. And I think one thing maybe good talk about is Home Depot gave a great presentation in Colorado, which I unfortunately didn't get to see. I'm definitely going to post a video of it, but, uh, they, they're, they're all in on Cloud Foundry. Uh, it's growing really rapidly at Home Depot in the presentation or even like, Hey, we ran out of capacity on our cluster, but that's a winning to us. Like when we realized that we needed to add a bunch of capacity, that was so a big moment for us because we're like, they've kind of gotten to the point where they realized that every instance that they run on, you know, CF is an order of magnitude or two orders of magnitude more efficient than any other system they have. Right. So what, whatever's there is goodness, but they were benchmarking themselves and they're like, we can now push apps through this system end to end and 15 seconds, but why does it take us four and a half hours to do deploys sometimes? Right. And what they found

was, is there were still people who wanted to kind of do manual checks or manual steps or tickets or something, even now that the process is completely transparent and automated. And this leads me to kind of our ambitions with, you know, what we want to do in the future releases, which is, you know, I don't know if you've seen, have you seen our new APM demo where the app performance real-time metrics and such? Yeah. Yeah. I think I, well, I, I think, yeah, I've seen that. You've also like demoed some good, uh, some good stuff on some cloud native after darks here and there. Yeah. I was playing around with the newest version of it for the first time on cloud native after the dark. Yep. You could tell cause I was learning about it. Um, and, uh, the thing I see in the future is, is that it, before we used to count on human judgment for better or for worse about these deployments. And then oftentimes even then we'd have 30 people or a lot of people ready to do the deploy, a major deploy so that if it needed to be rollback or firefighting, we could send humans into the blaze, uh, once more onto the breach, um, as, as they said in, uh, Henry V. And, uh, the, the, the thing for me is, is that Netflix doesn't do that. Like if you went and talked, there's an awesome article on Pivotal, um, you know, the collaboration between Pivotal Netflix. And while they did endorse spring boot and spring cloud at a, you know, kind of completely authoritative way, they're like, we run Netflix on spring boot and spring cloud for our group. They also said like, by the way, our Atlas metric system is the only way we go this fast because you can't go fast with your deploys. You don't have a metric system constantly telling you immediately in real time about how things are behaving and the ability to shape traffic out a little bit at a time. So I do think that if you think about these cultural transitions from kind of human judgment and big team and go slow and use a

lot of humans to go fast until we really start to perfect this metric system as well, people are still going to, even though the system can go fast, you still want that insight into it. So I think that's an important, important thing for us to, to continue to bring to market. And I'm excited about that, that, you know, alpha code that's there and it's going to be in the product very soon. Yeah. And I, and I, I think, I think that connects with another sort of, I don't know, message or, or, or philosophic take on all of this to that. I find myself and ourselves collectively having to get across to people is, is if you want to be a software producing company or good at software, one of the best practices of software companies is that they don't necessarily use just random off the shelf ways of monitoring and managing their stuff. Like they really do instrument and they, they need to have a very good, their applications have to have a good understanding of how you're monitoring it and looking at it. And so you should expect from, from vendors like us and the people who are producing the software to actually be providing you ways of monitoring and managing it. Right? Like, I mean, we, we have good partnerships with new relic and app dynamics and all sorts of other people who can plug in and monitor things, but you're going to need to spend a fair amount of time doing customized instrumentation because you're running that software. And it's, it's not just like the old days of just, you know, running some ping stuff and monitoring it. It becomes part of the core competency operating it. And so that stuff should probably be baked into the platform more and more. Well, I just, I, you know, I just want to change what a deploy looks like. And a deploy should be a share of traffic, not a hundred percent. So you should be learning about your new code in real time. You should have real time metrics about what's happening such that you don't have some sort of, you know, uh, Oh, I got to wait five minutes to see what our APM told me it has. Um, and, uh, you should be able to then automatically shave on more and more traffic. Now, part of this

gets into how you design your software too, in terms of backwards compatible APIs, you can roll the software out, roll it back if you need to. That's important. Um, but you know, I, I actually, you know, I think app dynamics, even though we're going to have the basics of APM and in the product, I think over time, there's always room for them to give even more advanced, you know, moon based metrics about what's happening. And I think I just want to make this very turnkey over time such that the deployment is always enriched by metrics. Cause I think that's just sort of like, you know, you flew here last night. You're not counting on a human anymore as much to fly your plane. You know, the, the computer is making judgments and I just think software can get to the same place where, Hey, this deploy is working or it's not should be a computer telling you, you know, how it's going. Yeah. And I, I think before, before we close out, you, you, you brought up the home Depot and I think it'd be good to go over, uh, you know, customer and momentum that we've had since last we talked to, to close out with, but that you, you're touching on another notion that I found myself talking with people a lot about where, um, to, to, to start at the end, to get to the beginning, they, they ask a lot of questions where, where annoyingly for them, the answer is, yeah, you don't really worry about that anymore. Right. And I, and I, and I think it gets to this notion you were, you know, metaphorizing to an airplane is with the right kind of cloud platform in place, you can actually start trusting the system a lot more than you used to trust it. Right. So you don't need that team of people there ready to put out the fire when you do something. And it, it takes a lot. I mean, we, you, you've been around since the beginning, we've been working on this platform for a long time to get to this point. Uh, but essentially once you have a trusted platform in place, there's all this part of your process and the way you're thinking about it, that turns out to be waste in the here and now things you don't have to do anymore. And then you can start to move faster and trust the system more. And I think, I think a lot of the hurdles that kind of get thrown in front of us before people are comfortable kind of getting into that environment has to do with, uh, establishing

and trusting that the system is actually going to run and operate correctly, which, which is a nice thing to get them to. Yeah. And I think that's where DevOps teams were kind of stranded for a little bit before they had a robust structured platform with these kinds of metrics, which is that they were seen as going faster, but not necessarily being safer. Exactly. That's right. People would use a word like cowboy in there a lot. Yeah. And, and in some ways maybe that was the necessary exploration phase, but at mega scale, you got to know that these, these big companies like Netflix, like they've got the thing instrumented to the ninth degree. Like, in fact, they even said that their cloud platform is really just a metric system in disguise. Like that's mostly what it is. Um, because, you know, to our point about cloud, um, and not about poetry, once you, uh, you know, clouds instead of scalable primitives, right. And I, that's what I believe. And it's scalable primitives for, for, for running software, um, and, uh, deliver often as a service, but the is the scalable primitives. So you have all these scalable primitives, which you can muster into existence, be it a VM, be it a container, be it load balancing, you know, in seconds. And the question is, well, how do you orchestrate all of that over time? And it's gotta be with data, right? Like you, you can't just have humans going, oh, that looks about right. Um, so I think that's a very interesting future is just like the degree of integrated metrics. Now that we've made the deployment and life cycle operations on the app boring, like that's great. But then how do you learn more and more? And then how does that app fundamentally drive an experience for someone on the other end is the ultimate reason it exists. So the performance and metrics of it is kind of the fun part, I think. Indeed. Actually, actually observing and seeing how your stuff is operating in production or in the wild is, I mean, that's, that's the, uh, that's the thing I try to emphasize to people is continuous delivery is great, but if you ignore that feedback loop, you're just sort of burning money or burning opportunity essentially, right? Like that's, you want to observe how things are working and what

people are doing with it so that you can improve the software, not just get uptime and have it running. And that, uh, that'll be the exciting thing to see panning out. And, and that gets to the last point, right? So I, I remember, uh, uh, you know, one of the more enjoyable aspects of the job is because of the type of, of customers we have, not all of them are, are P are companies that you deal with every day, but all of them are like regular real companies. And so I, I, I remember watching kind of our, um, our, our, our, our lead up to basically the Home Depot using us and doing that presentation you were talking about. And it's exciting. Cause like I go there every other, every weekend. And so it's, it's, it's, and, and you see this across all sorts of, uh, the types of customers that we have. And, and so, I mean, that's, that's one thing, uh, that I wanted to close out with is like, so how would, I mean, go over the, uh, the, the, the types of new customers we've been signing up and like what, what that's looking like at the moment, because I get asked this question a lot, being a former analyst, like people often ask me to characterize like what the market share is of Pivotal or like how we're doing out there. And, and so far there hasn't been a lot of great analysts writing up to like answer that question. So we're, it's kind of behooves us to define who's buying this and, and, and what that market looks like. And I think, I think you would have a, uh, a good take on that. Yeah. I mean, I think, I think that, uh, it's been really interesting to see how strong our growth has been in the fortune. I'd call it 200 and even 100, because if, if you think about the people that have an at scale problem, right, they're not like running an app, right? Like they want to deliver N number of apps. And they'd also are transitioning from a period where they were doing a lot of bespoke it of the kind we talked about before. So they have a very traditional it organization. Our platform offers a huge instant arbitrage for their ability to deliver custom software, especially when combined with something like spring boot. So like we come in and they're like, wow, so this is how we build and run custom software at scale.

Um, and you know, the other company I mentioned there was Comcast. I mean, that's another daily brand. And in fact, my fun game that I play with myself when I'm on the road is I look at, uh, names on buildings and I keep track of market share. Like how do I keep track of market share? Right. Like I'm, I'm, I'm pretty competitive. And so anytime I drive by, uh, in my, my, my quote is if you have, if you're big enough to have your name on a building, you're big enough for Cloud Foundry. Um, and so I kind of keep track of the market share of, you know, names I see on buildings versus how we're doing. And on my ride into the, you know, um, right into the hotel last night late from JFK, really remarkable number of the banks and major businesses of New York city that, um, you know, have their brands out, you know, as I drove by, I was like Pivotal Cloud Foundry customer, Pivotal Cloud Foundry customer, Pivotal Cloud Foundry customer. So it's getting better. It's getting pretty, pretty neat. Um, so, you know, there's a lot of momentum there. And, uh, especially when you think about, you know, the fortune 200, uh, and those kinds of customers. So, you know, I think specifically what's been happening though, is if I were to characterize it in some depth is, you know, banking has been facing an efficiency crisis. So the reason we've become pretty popular in banking is people realize that they used to be so successful. I don't know if you remember pre 2008, like how much, you know, banking people made and how lucrative a business was. Right. So they had all the money in the world to throw at every project. So guess what? Every project got its own it team, right? Like its own everything. And so they're coming from a, an era of vast heterogeneity, like very privileged heterogeneity in a sense. And they're looking at this and going like, well, you know, we're not being valued quite as much as we used to. Um, and we actually need to take some, some money out of this and we need to be efficient. And to some extent, they also want to digitally differentiate themselves and customer experiences too. But I think they're kind of, they're kind of like having

that hangover moment when the chicken walks by and, uh, they're like, wow, we could be a little bit more efficient here. Um, and so I think there's been some unbelievable traction in the top 10 banks in the world with Pivotal Cloud Foundry. In fact, that's how I'm spending my next two weeks seeing about six of them. Um, because of the need to drive digital consumer experience as well as get more efficient. And one of the coolest things that's happening is that we've got this one, two punch going on in banking where we're teaching their developers spring boot for all the new apps. Um, but then we also started to actually go through their Java estates and say, well, what could you run in a more of a spring boot fashion if we did a light migration? And it turns out a lot of the apps that were written with these heavyweight app servers actually only use the same, you know, kind of servlet interfaces that modern cloud native apps use, but they had too much extra boilerplate around them. So we're actually migrating with some of those for customers. And that's really exciting because that allows them to get kind of critical mass in the platform. Right. And then the third thing that we're developing in that case is that a lot of banking needs a lot of integration work. Um, and so we launched Tibco business works on Cloud Foundry. That's heavily used by a bunch of the big banks, um, so that their integration, you know, and enterprise service bus is, is actually able to be run in a cloud native way. And then some of those customers don't even like that pattern anymore. And now they're starting to use spring cloud data flow to enable kind of integration and, uh, you know, lightweight, uh, data microservices. So it's sort of like this one, two, three punch in those environments where we teach their developer spring boot. Um, we help them migrate apps that might not have needed all the heavyweight boilerplate that they have. And then we're also even starting to work on some of their integration use cases all on top of, you know, kind of Pivotal CF and frankly spring boot. So that's, that's kind of been fun. Um, and, uh, auto manufacturing is another place that highlight where, you know, you saw, we launched the Mercedes me app on Pivotal Cloud Foundry with spring boot. That was incredible. And I'd say similar to banking,

like, you know, some high percentage of the top 10 auto manufacturers are working with us. And the reason is they've got this whole connected car initiative hitting them. And they're going through exactly what you and I talked about, which is like, how do we become good at running software? Like really good. Yeah. Like there is a fight right now between Mercedes and BMW for their best connected car app. Like when you talk to those teams, it is like a hockey fight. Like it's a match and they're like, we're going to write a better app and we're going to have a better connected experience. So like it's become an existential part of their product development. So those are just two, you know, and then obviously in, in retail, we've seen a lot of growth with, uh, you know, um, both Kroger and home Depot have, uh, spoken about their adoption of, of Cloud Foundry. And, uh, when we're heavy at a SpringOne, when I was there learning about, you know, cloud native platforms, frameworks as well. So, you know, those industries each have their own challenges. You know, I think retail always wants efficiency, but now they're trying to do, you know, new product development and, uh, you know, delivery, you know, digital outreach, marketing, et cetera. So just those three industries, but those are a lot of the things you drive by when you go down the street, right? You're like, okay, car dealer, uh, bank, a big retailer. Those are the experiences we all have every day. Yeah. And, and, and, and for me personally, like what's, what's exciting about it is like, for the most part, the goals of all those companies is like making their software better, like the software I'll end up using. And so like it directly, uh, you know, benefits me essentially that, uh, that, that they can improve the, the, uh, you know, the way I interact with those companies. It'll, uh, it'll turn out to be quite nice essentially. Well, yeah. And most importantly, just to close on this, it gives me fiance credibility because when you work, that's right, when you work on complex, you know, in data center software, they, people have a hard time figuring out what you do, but she hears the calls I'm on and the names that I'm talking about a lot now. Yeah. And so we drive by and she's like, oh yeah, Home Depot. I've heard of that.

Exactly. It's, it's, it's, it's going to be a much more thrilling, uh, holiday season explaining to the relatives what, uh, what it is I do. Exactly. Exactly. Well, that's good stuff. I think, I think this is a good catching up around the, uh, the, the, the Pivotal Cloud Foundry 1.6 release. Like we hit upon some of the major items there and there's, there's plenty of, uh, we'll put in the show notes for this podcast. There's plenty of details about all the little and big stuff in there beyond the kind of, uh, context that James and I were talking about. And, um, so thanks for making the time there often. I, I always enjoy like the, uh, the, the Mad Men Manhattan background noise of, of New York, right? Just kind of floating around in the back there. It's fun to, to do this one just to close out, uh, overlooking, uh, Union Square Park and the people coming and going in New York. It's a, it's a, as a San Francisco person, it's still exotic enough for me to see. It's kind of fun. Exactly. Well, and, uh, so thanks for listening, everyone. You can find all the show notes and everything that we have. If you go to pivotal.io slash podcasts with an S or without an S and with that, we'll see everyone next time.

Tanzu Talk · 🍕 Get more pizza.