Exploring the Collision of AI and DevOps Worlds at KubeCon North America Day Zero

Hello and welcome to Data Driven, depending on how

you're consuming this. My name's Frank Limogno. You can catch me at Frank Digs Data

on most of the socials and franksworld.com,

datadriven.tv, and of course impactquantum.com. And

today was day zero, I guess, of

City, which is really nice fall weather, honestly, right now. Got my

badge. Qlik is a sponsor. So shout out to

my peeps at Qlik. I got a chance to walk through the expo

floor as folks were setting up and I noticed a very consistent

theme. And that theme is that a couple of things.

One, most data people really

don't interact with Kubernetes. And let me explain.

My brother from another mother, you know him as the co-host of Data Driven. He,

he's a, you know, he is the SQL Server genius. Yes.

And he's never really worked with Kubernetes. And I asked him, I was like, has

it come up at all in your, in your

career, your consulting or whatever? Has Kubernetes come up at all? And

apparently no, it has not. Which I find very intriguing

that somebody like Andy, who is, you know, definitely the

SSIS guru, SQL Server data warehousing guy, hasn't really

touched Kubernetes. AI people, of which I

am. I never really had to deal with Kubernetes. Actually, it wasn't until I

really appreciated, with the Red Hat, that I appreciated Kubernetes at all.

Even though realistically there's probably some form of Kubernetes underneath

Azure, all the Azure data services, which got me thinking

about how the data world

is going to interact with the

DevOps world in the future. For the majority of

its life, Kubernetes was really about DevOps, right? Delivering services at

scale. Well, what's going to happen when those

services that are going to be delivered at scale are going to be AI, AI

inference, right? That those 2 worlds are colliding.

That was really kind of the theme I saw. I saw a number of other

advertisements and kind of as people were setting up the booths, I always like to

check out the competition. Another theme was

their data, serve data through Kubernetes, serve whatever. And

another theme was token costs, right? Managing your token costs.

And basically token spend and things like that. So I

think it's interesting to see how

the DevOps world and the

AI world are forcing this conversation together. I don't know if the data

world is going to have this, although I did see a number of Postgres scaling

up on Kubernetes and there was another data platform too. So I find

that interesting that we're seeing kind of the Kubernetes world and the AI world

kind of merged. I saw this coming at Red Hat, but I

think the challenge is, and again, this is a plug for my

employer, the challenge is, do you really want your data scientists

messing around with your Kubernetes infrastructure? The

key thing to keep in mind here isn't so much

the infrastructure. I think people get wrapped around the axle about the infrastructure,

right? And it's not infrastructure for infrastructure's sake. You want to get these

POCs and you want to get them out to production, right? And I think that's

really where the rubber of the— is going to meet the road, needs to meet

the road over the next couple of years in terms of seeing what a good

return on investment is for AI, because that's also going to be a hot topic

in the not too distant future. So I think that what

gets me excited about working at Qlik is that I think we have that answer.

We call it the control plane for enterprise AI. Stop by booth

276, I believe. I saw the Qlik demo, I saw

ClearML, I saw what they were working with Qlik, and I was like, this is

it. This is the answer. Because you want to have the space for your

infrastructure, your admins. You want to have a safe place for the

AI builders to make, right? The data scientists, the AI engineers. And you

want to have a place where you can have people are using the apps that

the builders use, right? So you really have that 3 layers of

use cases. And I'm seeing a few other Kubernetes platforms

and they force AI engineers to learn

Kubernetes, which will they learn it? Maybe,

but realistically they shouldn't have to do it. And I wouldn't

trust your average data engineer. Sorry, Andy. Although you're well

above average data engineer, Andy, with anything that is related

to, to managing Kubernetes at all, because it's a very different

skillset and it is a different world. It's a different mind

space. There's different concerns. But we all have to work together

to make enterprise AI work. I do have some footage I took with the

Osmo, and I'm hoping to edit that down tonight

and get an early start tomorrow because we're going to have fun, fun,

fun. So far, I've been very excited about, about KubeCon.

So with that, I'm going to end the stream.