Scott Rosenberg on Unified Platforms, IDPs, and the Future of Kubernetes

Hey everyone. Welcome back to our day three coverage of Kubernete, of Kup Con here in Atlanta. It's been an exciting couple of days.

We've had a lot of guests, we've talked about a lot of things, but like any good conference, in person conference, the best conversations take place in the hallways. And I've had a lot of those too. And I'll be talking and writing about that in the days and weeks to come.

Let me introduce you though to our next guest. His name is Scott Rosenberg. Hey, Scott is with a company called Terra Sky.

Is that right? Yeah. Excellent.

Scott, welcome to Text Drunk tv. It's great to have you here, Ben. Thank you.

It's great to be here. Absolutely. So we'll talk about Terra Sky, we'll talk about Cube K, but let's take a moment talking about Scott, give people a sense of who you are, where you've been, like your journey.

Yeah. Yeah. So I grew up in Chicago, moved out to Israel, um, you know, back in 2005, and really started my tech career around 20 14, 20 15.

Started in the mainframe world. Really? You know, they still exist today.

Apparently. They, They know they still, they, and they will 20 years from now. Exactly.

Believe me, they're not going anywhere. It started there, moved into more the VMware, uh, you know, virtualization private cloud area. And then all of a sudden around 20 18, 20 19, just this bug of Kubernetes hit me.

Um, that's when I joined Terra Guy as well. Um, okay. And ever since I've been just working up in the Kubernetes face, I've been privileged to be a contributor to Kubernetes for six plus years.

Um, very cool. Working on a bunch of the other ecosystem tooling, cross plane backstage, um, and like what, what, what I'm known for, which is awesome, is I have never been into a single session at CubeCon that I'm not doing, because as you were saying, the hallway track are the most interesting conversations Absolutely. That there can possibly be.

And you get to meet all these amazing people here. I can watch the sessions live or the recorded sessions later on. Absolutely.

And people, I just wanna get to meet the people Because you only have that window. Right, exactly. And then everyone goes back into the world.

World And all these people I'm writing with on Slack every day, it's like you find Them, meet them in person. Well, once you meet. Right.

Oh, that, and that's a big thing too. You know what I find people I meet on Zoom, they're always taller in real life, but Exactly. That's have a Body.

Well, because it's, because you never, you know, it's hard to judge how tall someone is from here. Exactly. Right.

But anyway, you know, you mentioned Backstage and of course Backstage Con this year Yeah. Was a big success platform con here was also you presented a platform con here And a backstage Con Oh, you did both. I, I had four sessions this year at a cube cut.

So, exactly. But good for you. Yeah.

Well, why don't you share a little bit about what you, uh, presented on? Yeah. No.

So I, backstage Con was really about building out a platform with cross playing backstage, um, and trying to bring together the consumption layer and the operation side from cross playing and trying to fix that. But like, my favorite conversation, my favorite talk I gave this year was actually a platform con, um, which was with a amazing, uh, guy Shimmer m uh, from a company called Play Tika Uhhuh. Um, and they're a unicorn gaming company in Israel.

I, amazing company. And we've been working with them for probably like 10 years or so. Really?

Okay. 5. Oh really?

Okay. So back in 2016. And really it was, we talked about the journey that, you know, I've been going on with him together and how we have moved from a do it yourself chaos that happens with early adopters, um, into a real full CNCF bank like platform now, and all the benefits that it's brought them and what the challenges were.

Um, and it was an amazing conversation because it was the interweaving of the technology with the cultural elements and how it was able to be done, um, with CNCF projects. Love it. I love it.

Um, Scott, so at some point, it sounds like to me, you didn't say it, but you made the transition then to really platform engineering. Exactly. Right.

I, that's basically what we did, right? It's so much of this was a cultural change. Right.

Um, one of the challenges like that we always see is DevOps. One of the reasons that there's that line, DevOps is dead. Right.

Which is so, okay. It's a cute line. It's marketing.

Exactly. But why is that kind of true? Because DevOps never reached its full potential.

DevOps was supposed to be a culture and became the synonym for Jenkins, GitHub actions, cci, whatever it is, right? And Kubernetes, and that's no, DevOps was a culture. Those tools fit well into that DevOps culture.

What I love about platform engineering is that platform engineering really is an implementation way to make DevOps actually succeed. And the two are actually together. Platform engineering is a mechanism that allows us to actually bring DevOp a platform.

Fruition A platform. Right. Right.

And that's what a platform helps with. And really what we did with them was, I, this was a over a year project of assessments, of interviews, of gathering data, everything, data driven, and really coming and understanding the different pains of each team, and then going and building out this platform together with the teams to get their buy-in to really restructure things around and build that full platform out from the ground up. Got it.

Right. With the challenges of Brownfield and large scale and all of that Real life. Exactly.

Yeah. You know, know it, it's funny, Andrew Clay Schaeffer is one of the, not founders, but a big, a big name in the DevOps space. He always used to say, he still says, I guess the DevOps you get is the DevOps you deserve.

And Yeah. Part of that is, is because you're right. A good, the, the heart of DevOps was culture, right.

It was how we interact in a team. And for too many people, they used to pay lip service to culture. And they, what I call bag dive into the tech.

Exactly. Yeah. Yeah.

Culture, culture, culture. Jenkins. Exactly.

Culture, culture, culture, gi ups. But they never really, they never really invested in the culture. Exactly.

And I think the other thing is, if you look at DevOps that way, it's not meant, it's not meant to replace the Agile. No. It wasn't supposed to be from the beginning of time to the end of time.

Right. Right. I I think it needed platform engineering because it had to sit on something.

Right. Now you've got, so now the way I look at the world Yeah. You've got the platform, right?

So it's like on the first day God created the earth. Right? Right.

You got a platform. Exactly. That's one way of looking.

I'm probably a little sacrilegious, but all right. Well that's good. But you know, you got the earth now, now we can build on.

Right. We could build it on the second day. We created developers in code.

Exactly. And you, and, and, and we do it in Scrum and, and Agile. And that's the best way or monolith.

Right? Right. But we, we had this code or coders and then we had ops and, and you know, it was like, can and enable almost Exactly.

Very different DevOps tries to bring that can kain enable, if you will. Right. Not possible.

Kain enable together where we, we, um, you know it, but it it sits on top of the platform. Exactly. And Agile plays in there and DevOps plays in there and SRE is in there.

Oh yeah. And security and observability and all these things that we build on this beautiful Exactly. Earth, this beautiful platform.

Yeah. It's just, it's almost like we did it best backwards though, in that we Right. We brought the platform in after all these things were running around.

And, And I think the truth of matter is, and this is like one of the things I've been saying in like a bunch of the platform engineering, uh, different communities is that, you know, having a platform is not a new thing. No. We have had platforms for decades.

What we haven't had is a unified platform. Right. And that is the difference here.

Platform Discip engineering is the idea that we need to focus on the platform level instead of just having it as a given. And we have to unify that platform. Because what we had was a bunch of platforms in the past, and we did have platforms, but they were all disparate platforms.

Yes. And we had chaos. Well, one is now just let's bring those platforms together into one unified one and build all of those tools on top of it.

Right. Which is so many people saying like, oh yeah, we wanna build our first platform. I said, you have platforms, you wanna evolve into a unified platform.

Yep. Right. And that's the key, I think.

Well, once we get it built, then we could rest for a day. But anyway. Exactly.

Let, let's go. Let, I wanna move to IDP. Yeah, right, because you were back backstage con.

Right. You know, Just on that in IDP though, it's the, I will say the tech world can't figure out acronyms for the life of them. 'cause what I always say is you connect with your IDP to your IDP to visualize your IDP.

You have an identity provider right. To your Internal portal. So Your internal developer platform, right.

Which IDP are we talking about? You talking about I meant the internal developer, the portal platform portal and both. Exactly.

But this, you know, they used to say when Platform, when I first became aware of this whole platform engineering movement, you know, basically if you got a handle on Kubernetes, you, you had a platform. Right, right. And the engineer was doing that.

I think that we've seen that focus change from, you know, handling your Kubernetes cloud native sort of architecture to, to IDPs. Yeah. And, you know, and platform portal there.

Uh, and it's interesting because it's, it's, some may say, well, it defo us from working, you know, perfecting Kubernetes. We got enough people working here Exactly. To perfect Kubernetes.

It's okay. It's, it's not going to exactly be an orphan. Um, but, but, but we are seeing that.

But here's another thing. You know what they say? Two people, three opinions.

Yeah. And we're starting now to see, uh, differences of opinion different communities. Yeah.

org. They have a, a booth down the hall from me here. Yeah.

A couple rows back. I work very closely with Luca and the community. Yeah.

We saw Platform Con here put on by CNCF. They offer certifications. Those people offer certifications.

Is this town big enough? Uh, I think it is. I think that, you know, obviously there's always gonna be some challenges, right.

In differing opinions. But I think it's very similar to what we saw in the GI UPS world happen when you had weaveworks that had Flux. Right.

Right. Flux. Then you have Intuit that go and do Argo ccb, and then you come out and Flux V two comes out.

Right. And you have these two tools. You even have the other ones like on the side.

Yeah, no, they're are, or Fleas and the carve cap controller, and you have some others. We see it in service mesh too. Right.

But what ended up happening, they all come, right? They go in their own directions. They're figuring out what GI Ops is, and then what happens, the Flux guys and the Argo guys come together and create the open GI ops initiative, which comes says, great, the, this is what GI Ops means.

We're different implementations of the same thing. And that's fine. Choice is Good.

Exactly. And I think that this is healthy because having two organizations, having two groups trying to figure out and really distill down what platform engineering really is and what its core tenets are, is so critical. Because once that happens, the two can come together, let everyone go off on their own approach.

Similar to how people with agents, they run twice an agent, whether they're doing a no, absolutely. Let's see what we get and try it out. But variety's the spice of life.

Exactly. And so, and I think, I think that's what Open Source is about too. It's about choice.

It is, it's about freedom to, to decide, you know? And and if and if, and if Backstage the code doesn't work exactly the way I want it to, I'm free to change that too. E Exactly.

It's right. And I like Backstage is one of the best, I think proves of like what a real, like open source community is looking like. Because what we've seen is like Spotify, when they built it initially and they contributed top stream, they work in a very specific way.

And then you had companies like VMware that came in, right. And they wanted to do something a bit different. And absolutely Red Hat came in and you see these companies have Different opinions, but it didn't stop Spotify from saying, you know what?

We wanna sell a commercial version of it, which is awesome, our vision, and it's all good. It is big enough for Everyone and everyone can push it in their Direction. That brings up another, another question though I want to ask you.

Yeah. I think one of the questions for the platform engineering community is, do you want commercial IDPs like the Spotify backstage? Which kind of, it gives it to you in a nice package.

It's all wrapped up nice. It looks good, it smells nice. Here you are package, just add water.

Or do you want me to give you sort of that old time open source, here's all your packages, you pick what you want, put it together and it's a DYI kind of thing. Right? So, you know, it's a really great question.

And it's like, you know, it's something that we've been grappling with for a while with a bunch of our customers and the people that I'm talking to. And I think it really just comes down to there's no one size fits all. And it depends on the customer.

Yeah. You can't be a jack of all trades. No.

And therefore, in a company that's large enough, right? A big enough unicorn style company that has the ability to really go and fine tune that internal developer platform to be exactly how they want and can base it off of open source and can have a large enough DevOps team to maintain something like that, that's a very viable solution that's probably gonna give them a better platform long term. Because there's a reason it's called an internal developer platform, because typically it's built internally.

Internally, right? Now that doesn't mean that on specific components that you don't have the expertise on, you shouldn't buy an enterprise version of it. Right.

I think that, on the other hand, platform engineering can help so much in the smaller companies as well. Right. And there, I think that you have to go with these commercial offerings.

They're prepackaged because A, you just, they Don't have the resources. They don't have the resources and they don't hit the challenges No. That these platforms can't cover and Scale and stuff like That.

Yeah, I agree. If you Have 50,000 nodes in Kubernetes, I promise you that most of the paid solutions out there ain't managing it. Well, no, you're gonna need to figure it out yourself.

Agreed. But If you have 50 nodes in your environment, I promise, fine. Any of those IDPs out there are gonna be fine to help you.

So Agreed. Hey, man, I appreciate it. Thanks for coming on here.

Awesome. This was great. Scott, if people wanna maybe follow you or Yeah, stay on top.

What's your best website? cloud, uh, you can check me on weekly, uh, podcast together with Victor Farik on, uh, the DevOps school Kit. Victor's a great friend.

Tell Victor I said hi. I will. I'm a good friend, a good old friend of Victor.

Victor's great. He exactly. Great Guy.

Uh, so we do an ask me anything, ask us anything on every Thursday, basically. Very cool. Um, and reach me out on Kubernetes Slack, CCF Slack, cross Plaine, slack, LinkedIn, any of them.

Exactly. All over. Love the community and great being here with you.

Thank you, Scott. We'll come back again. Hey, we're live.

We got more coming to you. And as we wrap up our, uh, day three coverage here of CubeCon Con. Native Con is Alan Hummel.

Stand by.