AI and Build Cloud with Docker’s Scott Johnston at KubeCon Paris 2024

This is Techstrong tv. Hey everyone, I hope you've been enjoying our coverage here. Live at Techstrong, uh, at Techstrong at Cube Co.

We are tech strong, but it's, it's, it's a little crazy. I mean, Scott, I don't, you've been to other Cube Con? Yeah, I Have.

I have. And this one is off the charts. You're Right.

Absolutely. I, I was with, uh, o Neil Gupta, who's the chairman of CNCF. Sure, sure.

And he told me last night, this is the biggest cube con bigger than San Diego. Yeah. If you remember San Diego before, Before Covid Yeah.

Yeah. Was the biggest swan. This is now eclipse.

That, so pretty cool stuff. Uh, you heard me say Scott, let me introduce you to Scott Johnston. Scott is the CEO of Docker and has been now for, how long has it been?

Eight years. Years. Just over four years.

Years. Four years. Been at Docker 10 years.

So I think I remember when you joined Docker, believe it or not, February, 2014. Docker is 11 years old tomorrow. Really?

So Solomon walked on the stage at Picon Picon 20 13, 11 years ago tomorrow. Okay. And brought Docker to the world.

So here we go. Very Cool. Very cool.

I mean, this didn't exist. This didn't give this 10 years ago, Right? No, it didn't.

No, this certainly didn't. com, I started in March of 2014. Okay.

So it was all ascending The time. So it was years. Yeah.

Wow. Crazy. Crazy.

Hard to believe. But there's been a lot of changes since then, right? Yes, yes.

You know, back then Docker was all about the container. Mm-Hmm mm-Hmm. Of course.

A lot of water under that container. Um, as they say, A lot of water under the container. That's good.

Yep. That's good. And, uh, Docker has grown, morphed, stretch, pulled, changed, come out the other side.

Yeah. With a, a very different business. A lineup That's Right.

Than than containers. That's right. Let's start there if you don't mind, Scott, at all.

Tell, share with the Audience. So, so our journey up to 2019 was really focused on the production use case. Yep.

Helping operators run containers at scale in the data center in the cloud across multiple nodes. And so of course, that involved orchestration. And back in the day we had an orchestrator, Kubernetes then came on the scene.

We eventually embraced Kubernetes, so on and so forth. In 2019, we realized that there was another opportunity for the company, and that was the developer market. Yes.

And so we did a massive pivot in 2019. Basically shed the operator business, shed the orchestration business to focus just on the needs of developer and specifically, how can we help developers rapidly as a team develop more secure higher quality software. And that's the journey we've been on the last four years.

And you know what it it, as you said, it was a pretty radical pivot. Big shift. But in hindsight, genius.

Right. I'd rather be lucky than smart, as they say. Right.

It's good to be both. It's Rather be lucky than smart. That's the story of My life.

But, but developer experience is now A key piece of it. Ab Absolutely. Absolutely.

So the, you know, we're going to get into some of this now though, but I wanted to, now you guys made some announcements recently. We did. We Did.

Uh, well, let's go there if you don't mind. Sure, Sure. So in this focus on developer experience, it's specifically on what we call the inner loop.

And that is what the developer's doing locally when they edit the code, build the code, test the code, verify the code, debug the code, and they go around and around that loop before they do the get commit. Right? So it's all, which is then the outer loop.

Okay. So how do you help 'em go fast in that inner loop? And one of the realizations we came to is seeing the data coming back, is that there's some points of that inner loop that are not efficient, that are not optimized where the developer is waiting.

So for example, the developer can wait in aggregate sometimes up to an hour a day for their builds to complete locally. They, they type docker build and they go get a cup of coffee or they go to lunch. And we came to the realization of like, well, wait a minute.

We can bring the power of cloud to that local build experience so they have the same developer experience, but offload the build to the public cloud. We're seeing 39 times speedups. Which to put that in context, that an hour becomes a minute and a half.

That's free time. Absolutely. That's time back in the developer's day To do other things.

You know, I've seen a, you've probably seen them too, a lot of these surveys that say developers only spend about 30% of their time developing Actually writing code. Yeah. Writing code.

Right. Right. 'cause a lot of it is waiting for sh Stuff.

Waiting for stuff Yeah. To render. Right.

Right. That's right. And, um, so give 'em more time back in their day to stay in the flow state, be creative, write code, because What do developers like to do?

Write code. They like to write cove. They're creative.

They want to build. Right. I agree with you, man.

That's where it's at. Um, so working on that inter loop, what else do we got going on? That's a big one.

So similar to the inner loop, uh, sorry. Similar to build, another big activity in the inter loop is test. Right.

Okay. So now I'm gonna, I build my container. Now I want to test it locally, but if that container, or if that, uh, application has like 20 different services that can weigh down the laptop or if they're developing on an M1 arm, but their production is X 86, now they have an architecture difference.

Yep. Right. So we bought a company in December called test containers, which allows you to have the same test experience locally as you have in the cloud cloud on these same principles of like, burst out to the cloud to do what the cloud does best.

And so we announced this week a partnership with Red Hat Beautiful. Where developer can develop locally, but if they wanna burst out to their Red Hat OpenShift cluster to run the test there, seamless one command deploy. Done.

So lucky and smart, I'll say, I'll take, I'll take lucky. I'll take lucky. Why didn't someone think of this earlier?

I mean, OpenShift and our big customers and big banks, it is what platform engineers use to deploy to. Yeah. They wanna bring, uh, works on my, they wanna avoid works on my machines.

They wanna make, they wanna make the, the test environment as close as possible to the production environment. And we're like, fantastic. Let's burst from the local test environment to the OpenShift environment.

I mean, call it good. So I'm, I'm not as technical as you probably, right. But look, one of the first big uses not a tech on when I was at my last company, when the cloud first came on the scene was what a difference.

Game changer for testing. Yes. Testing was a expensive time consuming Yep.

Pain in the butt. A hundred percent. Now all of a sudden I had infinite, not infinite, but near infinite scale for my unit test.

That's right. And my, all my, you know, load testing and all of that, we already, so testers already knew that the cloud was your path to buying, uh, you know, a hundred thousand different flavors of laptop and everything. Right.

Right. Why, why did it take so long? You think to do this with containers?

You know, I think, I think at the time, back in the day, right, you had a dedicated test team, right? Remember this and it was, it was very much waterfall. So like, write the code, someone else built it, someone else tested it and like waterfall down the way.

And in those days, okay, you deploy once a month, you deploy once a A once a quarter, twice a year. Once A year. Yeah.

Yeah. Exactly. But now we're in this world where like value shipped is the premium.

Yeah. So how quickly can you ship value to production? Yep.

And so that combined with automation, combined with, again, humbly the ease with which the containers become that unit of work Yep. Is pulling more and more that into the developer environment. So how do you make it easy and automated for the developer to do testing as quickly as possible then versus throw it over the wall to someone downstream.

So I think that's the dynamic now. And is that like, hey, the faster we can solve issues, the better that quality, better quality software that's gonna be developer can take action and then move on. I mean, I like it not only That you have a, you have a better test coverage.

Yes. Yes. Exactly.

I I liken it to a manufacturing example, which the, the Japanese manufacturers brought, which is they have this notion of the Andon cord on the line, right? So if a worker sees a problem, they pull that Andon cord because they know fixing on the line costs a dollar. You fix an inventory costs a $10 Recall costs you a thousand.

Fix Recall costs you a million bucks, right? Yeah. And so very much what we can do from manufacturing, we can bring to software.

Absolutely. Like help the developer solve it. Right Then.

Not in ci, not in production, not Oh my God. When the customer sees it. Yep.

Right? And so that's what we're doing. We're helping developers.

It's a Shift left. Solve, solve problems, solving the problem. That's what it's what you're Doing.

Solve problems as soon as Possible. Stuff. Both of these are available now or or just announced now.

Variable. Now. Variable now.

Very cool. So the one other thing we're doing this week, and you're gonna hear a common theme here, which is again, the hybrid, local and cloud. So of, of course the meme of the moment, genai, right?

So Genai, uh, works fantastic on laptops that have A GPU, but those, those laptops have a fixed capacity of GPUs, fixed speed of GPUs. So what we're also highlighting this week is the ability to burst that LOM out to a cloud cloud and run it on a cloud node with a big beefy GPUA real Nvidia deal Or, or, or tens of Nvidia GPUs in the cloud. Right?

So again, how do you speed up that iteration, take advantage of the cloud for what cloud is good for? Right? I mean, look under, it's that same mindset as as moving your testing off, right?

That's right. Or your builds Up and this is a great use for the cloud, right? It's because it is so many people, I wanna host my app in the cloud.

I wanna, I need good identity, you know, security in the cloud. But the cloud works best for I think little jobs like this very Specific Yeah. Surgical boom Boom in out of all that's right.

You know, I need what I need and I'm done. I'll use it again next time. Versus these sprawling infrastructures that, that we spin up.

I mean, Production has found ways where spiky workloads work well in the cloud and retail, retail, you know, Christmas season or Black Friday, like, okay, great. But we also realize that like there's so much power there that the developer is not able to harness. 'cause it can be difficult to set up and provision and secure.

So we, our brand is simplifying, right? Right. So if we could simplify all that, actually we take it completely off the developer's plate.

Developers just use the same commands behind the scenes burst out to the Cloud. And that, that's, look, like I said, sometimes I sit around and say, why didn't I think? 'cause Well, There you go.

There you go. Exactly. Hey, I wanna bring up another topic.

Yeah. com, we're cloud native now and all these other That's right. Well, they can't see it on my, but that's over here is our real background with all of our sites.

At the heart of a lot of this is that whole CICD pipeline process. Mm-Hmm. And it's, you know, we're doing this big research project right now called DevOps next, right?

Where we're, we're looking at, we, we don't, we're moving away from cobbling together point solutions and it's, it's a natural evolution, right? Sure. To more of a platform, more of a, you know, it, it it organic if you will play.

We're not just cobbling. Right? Right.

And the heart of that is that CICD process. We've come a long way, but there's a long way yet to go. That's right.

That can be done. That's right. A lot of these little things point kind of things that you're talking about fit into a larger picture of how do we speed up CICD.

That's right. That's right. And, and, And so what's the docker view there?

And, and look, CICD plays a really critical role for, uh, particularly that intersection between dev and ops of like, what are the, what are the final checks that this app has to go through, particularly for regulated industries? Oh yeah. Mission critical workloads.

And there's a set of tests and certifications and checks that absolutely have to be done before that workload goes into production. But we hear from customers that sometimes it can take them 5, 6, 7 times looping through CI before they actually go to Able get, Get, get, yeah. crc.

Well, He get, well, he gets bounced before It gets into production. Right. And that, that means latency, that means time, that means the dev, the dev is waiting for stuff to come Back.

That's 30% number. Right? That's right.

That it's, that's exactly right. 'cause they're waiting for something else to finish before they can like, take action. And so we see that and we see what we've been bringing to market of like, okay, if we can actually help the dev solve it earlier, maybe instead of five, six, or seven times through ci, maybe they go once or twice through ci.

And so everyone benefits faster delivery of value. The dev is solving problems right then and there, not 30 minutes later, not 60 minutes later when it comes back from ci. And the CI team is known for a high velocity enabler.

So everything, everyone kind of wins in bringing absolutely these best of both worlds. Absolutely. Best of inner loop to the best of outer loop, as we call it.

Docker's not done with this stuff. We're not done 11 years going, we're just getting going. We're just entering our second decade.

All of us, all of us here. Right. Second decade, You're a hundred percent cur, you're dead on on that.

Right. And I, again, I think about Wow. Wow.

Just, wow. Do you envision a time where instead of helping these companies with CICD kind of accelerators or catalyst people go to DACA for CICD? It's an interesting question and it it gets into company evolution question as well, right?

Um, I mean, we have found a lot of success the last four years, as you referenced, staying focused on the needs of developers. And I think what that could mean with automation and where a lot of the technology is going is, does, does it still look like traditional Dev CICD production five years from now, six years from now? I don't think so.

It could, it could radically change. And and it's not only the result of automation, of new technologies, but it's also now we're injecting, um, gen AI and data into the equation as well, which now has to be part that Area. Not just gen ai, but I I say ml s All up.

You're right. Yeah. All up.

And, and that's fascinating too, right? 'cause now you're shipping not just code, you're shipping models and the data with that model, and you're having to go through iterations that today we go through with code, that you're gonna be iterating with models, iterating with data that They're gonna Right. They're gonna iterate themselves almost.

So, Yeah. I mean, you have to automate it, otherwise everything's gonna slow down again. Right?

And, and now you have this new persona, the data scientist who is upstream helping tune the model, but then it's the devs that is taking that and incorporating it. So, so we're gonna evolve. We're gonna evolve.

Sorry. We as an industry are gonna evolve. Yes.

And Docker's gonna be there. It's Gonna evolve right along with it. Docker's gonna be there to serve that development team and help them take advantage of all these great technologies and the data and the, and the models to make great apps.

I love it, Scott. All of the things we're taught, well, what we just spoke about's not available today, but everything else we spoke about is Yep. com.

com. That's right. All right.

Hey man, thank you so much. Always good To talk, Alan. Alright.

Scott Johnston, CEO at Docker. Lots of stuff going on there. This ain't containers anymore.

Check it out. We're live in Paris. We'll be back in a moment.

I think we have tenable up next. Stay tuned.