Liran Haimovitch, Rookout | KubeCon + CloudNativeCon Europe 2023

This is texturing TV. Hey everyone. We're back here live.

Our continuing coverage of kubecon here in Amsterdam. We have 10,000 people in attendance at this coupon crazy, which is crazy is Right sold out amazing event. I hope you've been enjoying our coverage all morning with Mike Mike pizzard our chief content officer on myself.

I'm Alan shiml our next guest. You know what I see him. Only a conferences he's ever comes to visit.

They're like a Jewish mother here. You never call but my friendly I'm I'm like, I'm a b****. Excuse me.

He's the CEO founder of workout, right? Yeah. So why I did say your name correctly for us on time of each.

There we go. So they're on I wanted to jump into as you call it the fourth pillar of observerability. But before we do just in case there is some people out here who don't know from workout.

They don't know who workout is. Why don't you give him a little bit of so real flavor rocket is all about bringing observability to developers. We think that you know observability started as a branch of monitoring which was very it and Ops focused.

And today the modern nature of observability is still very operations focused dates sris, and not Ikea administrators, but still they are asking different questions and they have different needs and the average developers and we think that it's part of the shift left. It's part of the fact that we want developers to totally focus on production own production on the customer experience. They need their own set of tools that are tailored for their needs to see what's going on in production.

So that's that's what brokerage is about building that platform. Absolutely and workout is look. I think I remember when you first came out it's about four or five years.

Yeah. So and you know, this is your these are your people this is your crowd right here at the cloud native world now. Probably one of the you know, we announced this morning around that hour what used to be container Journal.

We've renamed now Cloud native now. Congratulations. Thank you.

But part of the reason look when I first started container Journal, I almost named the docker something. I'm glad I did it but even the word containers The cloud native movement is kind of outgrown just containers or even just kubernetes and hence Cloud native. Now.

One of the biggest things is observability right in the cloud native. universe and you know you have some unique. Not only they're not unique, but some well thought out.

Ideas around observability as I mentioned earlier. We're going to talk about the fourth. Sort of pillar, but before we start the fourth, you should tell us what the first three are.

So I think most of the audience is familiar with them. But now the core the first let's start the first and Fogo the first is logging I mean logging is been fast for decades maybe even a century by now every hello world application anybody's ever wrote is a foam flogging and that's what we all started with. And then later on we learned that logging don't scale all that whale and we started with metrics which were popularized by Google as part of the S3 and books and got them to build a planet scale software for the first time.

And after that we were said spans which helps with Cloud native and micro services and all that distributed tracing full of things, but both metrics and Tracing Our zoom out approach. It's about how do I get logs and look at them over time over space? How do I look at Logistics between microservices?

How do I get the big picture and so on which are exactly the questions srees are asking the questions if there is should be asking. But most developers don't spend most of their time zooming out. They spend most of their time zooming in zooming in a specific bugs zooming in on specific functions.

They have to make more efficient or refactor zooming in features that I need to add or remove and for that zoom in mechanism. We didn't used to have a good solution for the past many years now. Absolutely.

Okay. So let's get to this fourth pillar if you will and and what is workout doing with it. So if you think about look at the you know, those damn log lines just I got here.

That's the you know, most common logline Engineers ever write special debugging Engineers that when Engineers use logline to understand the state of the application. Where did we go to what are the variable values? How did we get here and so on and so forth, but the thing is for logs, you have to be very prescriptive everything you want in the log.

You have to write you want a specific variable. You have to write if you want the dictionary. You have to write it you have to think about how do I stringify?

How do I manage the size? How do I make sure not lose data on the one hand? How do I make sure the local itself is not causing harm.

It's not too expensive to ride in performance. It's CPU. It's not creating honestly input or output and so on and so forth.

It's really a hard balanced strike. And which is why most loglines end up being very like cluster and not very valuable what most Engineers would really want is just saying tell me what happened here. And that's what the snapshot is all about.

Tell me what happened here in this line. Get me all the variable values get me the entire stock price get me this the version of the source code everything that happens and that's what Snapchat are about now. If you think back to it, we had this in one form or another for a few decades now, especially, you know yet looks at Cold arms when the process crash.

Let's give you a code of which exactly that the state of the application it's last second of life. But Snapchat is about bringing down to the cloud native world to distributed world. It's about making that scalable and easily usable.

Excellent. So I've just serve, you know, Lauren they don't see because they only see this background. They don't realize we're sitting here is in this.

Like an airplane Hangar with the sun coming down on us with 10,000 people walking by walking by across from us is the Rancher by suits and they're kicking soccer balls on the target. It's a conference. It's fun.

It's a lot of fun. It's good to be back at a conference like this. so if I'm not mistaken the last time you and I spoke was Detroit.

I think we spoke in the other devops. All right. Yalla, devops was about last September.

Yeah. You had nine months ago. tonight we spoke a little about snapshot stands both a little bit about shift left versus shift, right how we want Engineers not just shift left is not so much about bringing the real world into the lab as it is about bringing developers out the real world because there's only so much you can get into the lab.

and I think one of the key aspects we've learned the truck out over the past nine months is that we 've always considered the key value proposition of workout did the agility the fact that we can get you anything anytime and you don't have to pre-specified. And over those nine months we've learned it's not just about the agility. It's also about the format.

We bring the data. It's about a verbosity of the data and we found that many of our customers and users don't care so much about the agility as they do about the Fidelity and the fullness of the data. And even if we run Burton the same data in a very static approach, it would provide just as much if not more value for some of the use cases.

And so while we're still big believer in big proponents is being able to determine in real time what you want and get it just like that. We've also learned that there's a huge value in snapshots in the fact that you can get the entire picture and I quote that, you know, just take a picture. It was a thousand worlds Snapchat is also thousand localized because it's gonna take us that much effort to accurately describe.

The state of the application is in traditional logging. so look one of the since then today. you can't talk more than 10 minutes without AI generate of AI not generative AI what?

What do you think? ai's role here might be I think it's gonna make Eugene impact. I think you know, everybody's talking about shed GPT and a general purpose AI.

I think we're seeing more and more efforts to integrate AI with domain specific data. I mean, it's not just about having AI being able to read code. It's about every AI being able to read code.

Well seeing architecture correlate the source code itself the static source code with Dynamic observability data kind of making those deductions automatically or at least pointing people in the right direction things is gonna be huge Boom for that and I think that once you've put in AI into the mix and why we will also working on automations, then the fact that you can dynamically decide what to collect becomes even more valuable think about it. You have somebody else summer. It's getting fed both with your source code.

You're observability data and it can shift it can shift your ability data and real time on the fly without writing code without really deploying and that's a key element of what we're focusing on being able to allow AI or other more traditional approaches to automating observability using our Dynamic instrumentation capabilities I love it. last question You know, we're hearing about hard times intact layoffs. Budgets being cut I'd say looking around at the floor here.

I don't see. Wondering what your impressions are. They just like in the truth is two sides.

I mean on one hand we have people. There's as far as I know in most countries in most roles in the tech. There is still more open positions than a Then people are looking for a job.

Yeah, it's it's a bit harder to look for a job. It takes a bit longer employers can be more picky. But still there are plenty of open open positions open.

And yet no people are walking harder. People are more worried about their jobs selleries are not going up every two months like they used to and especially when it comes to new expenses new hiring new vendors companies are more deliberate. And I know that walking with our customer base many of them have found that logs are a huge expense.

And in fact most the expense around blogs is formal driven. It's around fear 90 some case 99% of the logs are never looked at. They're very long say huge expense sure and it's all about being unable to collect real data when you need it and we're seeing more and more customers prefer to take more Dynamic approach to absorbability rather than trying to collect everything hold everything.

Costs of Fortune you can most companies between not to pay that today and just keep what you need must keep some of what you must maybe it's cold storage and not hot storage and rely much more on Dynamic approaches to observability to get what you need when you need it. Actually your life we got for people want to get more information on workout. We are where do they go?

So feel free to reach out to us. com. And if you happen to be it Cuban itself, we have a booth right here where we'll be happy to host you and show you cook demo.

Excellent. All right. We are live in Amsterdam.

In the sun. Hopefully that'll pass. I love those on you.

Love the conference is a huge black. Oh, you're right cover. You're not sure you're going in you can spend two day in the conference himself seeing the sun.

No, the sun is right. There. Here comes inside much live here.

All right be well you are we'll be right back with our next LeRon. We'll be back with our next guest. In a second.