 From around the globe, it's theCUBE with digital coverage of IBM Think 2021, brought to you by IBM. Well, good to have you here on theCUBE. We continue our conversations here as part of the IBM Think Initiative. I'm John Waltz, your host here on theCUBE. Joined today by Marco Novakovic, who is the co-founder and CEO of Instana, which is an IBM company specialized in enterprise observability for cloud native applications. And Marco joins us all the way from Germany, near Cologne, Germany. And Marco, good to see you today. How are you doing? I'm good, hi John, nice to be here. You bet, yeah. Thank you for taking the time today. Well, first off, let's just give some definitions here. Enterprise observability. What is that? What are we talking about here? Yeah, so observability is basically the next generation of monitoring, which means it provides data from a system, from an application to the outside, so that people from the outside can basically judge what's happening inside of an application. So think about you're a big e-commerce provider and you have your shop application and it doesn't work, observability gives you the ability to really deep dive and see all the relevant metrics, logs and application flows to understand why something is not working as you would expect. So if I'm, or just listening to this, I think, okay, I'm monitoring my applications already, right? I've got APM in force and I kind of know what things are going on, what's happening, where the hiccups are, all that. How, what is the enhancement here then in terms of observability? Taking, it sounds like you're kind of taking APM to a much higher level. Absolutely. I mean, that's essentially how you can think about it and we see three things that really make us and Stana and enterprise observability different and number one is automation. So the way we gather this information is fully automated. So you don't have to configure anything, we get inside of your code, we analyze the flow of the application, we get the arrows, the logs and the metrics fully automatic. And the second is getting context. One of the problems with monitoring is if you have all these monitoring data silos. So you have metrics on the one side, logs in the different tool. What we build is a real context. So we tie those data automatically together so that you get real information out of all the data. And the third is that we provide actions. So basically we use AI to figure out what the problem is and then automate things. Is it a problem resolution restarting a container or resizing your cloud? That's what we suggest automatically out of all the context and data that we gathered. So you're talking about automation, context, intelligence. You combine all that into one big bundle here that basically that's a big bundle, right? I'm not gonna, a giant vacuum if you will you're ingesting all this information, you're looking for performance metrics so you're trying to find problems. What's the complexity of tying all that together instead of keeping those functions separate? And what's the benefit to having all that kind of under one roof then? Yeah, so from a complexity point of view for the end customer, it's really easy because we do it automated. For us as a vendor building this, it's super complex. But we wanted to make it very easy for the user. And I would say the benefit is that you get, we call it the meantime to repair, like the time from a problem to resolve the problem gets significantly reduced because normally you have to do that correlation of data manually. And now with that context, you get this automated by a machine and we even suggest you these intelligent actions to fix the problem. So, I'm sorry, go ahead. Yeah, and by the way, one of the things why IBM acquired us and why we are so excited working together with IBM is the combination of that functionality with something like what's an AI ops. Because as I said, we are suggesting an action and the next step is really fully automating this action with something like what's an AI ops and the automation functionality that IBM has so that the end user not only gets the information what to do, the machine even does and fix the problem automatically. Well, and I'm wondering too, just about the kind of the volume that we're dealing with these days in terms of software capabilities and data, you've got obviously a lot more inputs, right? A lot more interaction going on, a lot more capabilities. You got apps, they're kind of broken down the microservices now. So, I mean, you've got, you got a lot more action basically, right? You got a lot more going on and what's the challenge to not only keeping up with that but also building for the future, for building for different kinds of capabilities and different kinds of interactions that maybe we can't even predict right now. Absolutely, yeah. So, I'm 20 years in that space. And when I started, as you said, it was a very simple system, right? You had an application server like WebSphere, maybe a DB2 database, so that was your applications. Like today, applications are broken down in hundreds of little services that communicate with each other. And you can imagine if something breaks down in a system where you have two or three components, it's maybe not easy, but it's handled by a human to figure out what the problem is. If you have a thousand pieces that are somehow interconnected and something is broken, it is really hard to figure that out. And that's essentially the problem that we had to solve with the contacts, with the automation, with AI, to figure out how all these things are tied together and then analyze automatically for the user where issues are happening. And by the way, that's also when you look into the future, I think things will get more and more complicated. You can see now that people break down from microservice into functions, we get more serverless, we get more into a hybrid cloud environment where you operate on-premise and in multiple clouds. So things get more complex, not less complex from an architectural perspective. You bring up clouds too, is this agnostic? I mean, or do you work with an exclusive cloud provider or are you open for business basically? We are open for business, but we have to support the different cloud technologies. So we support all the big public cloud vendors from IBM to Amazon, Google, Microsoft. But on the other hand, we see with enterprises, maybe there is 10, 20% of the workload in the public cloud, but the rest is still on-premises and there's also a lot of legacy. So you have to bring all this together in one view and in one context. And that's one of the things we do. We not only support the modern cloud native applications, we also support the legacy on-premise world so that we can bring that together. And that helps customer to migrate, right? Because if they understand the workload in the on-premise world, it's easier to transform that into a cloud native world, but it also gives an end-to-end view from the end user to, we always say from mobile to mainframe, right? From a mobile app down to the mainframe application, we can give you an end-to-end view. Yeah, you talk about legacy in this case, and maybe cloud services that people use, but a lot of these legacy applications, right, too, that are running that are still very useful and still highly functional, but at some point they're not going to be. So would it be easier for you or what do you do in terms of talking with your clients in terms of what do they leave behind? What do they bring with them? How, what kind of transition timeframe should they be thinking about? Because I don't think you wanna be supporting forever, right? I mean, you wanna be evolving into newer, more efficient services and solutions, and so you've gotta bring them along too, I would think, right? Yeah, but to be really honest, I think there are two ways of thinking. One is as a vendor, you would love to support only the new technologies and don't have to support all the legacy technologies, but on the other hand, the reality is, especially in bigger enterprises, you will find everything in every version, right? And so if you want to give a holistic deep view into the application stacks, you have to support also the older legacy parts because they are part of the business critical systems of the customer. And yes, we suggest to upgrade and go into a cloud native world, but being realistic, I think for the next decade, we will have to live with a world where you have legacy and new things working together. I think that's just the reality. And in 10 years, what is new today is legacy then, right? So we will always live in a kind of hybrid world between legacy and new things. Yeah, you've got this technological continuum going on, right? That what's new and shiny today is going to be old hat in five years, but that's the beauty of it all, obviously. And talk about AI ops. I mean, let's go into that relationship a little bit if you want to me, eventually what is observability set you up to do in terms of your artificial intelligence operations and what are the capabilities now that you're providing in terms of the observability solutions that AI ops can benefit from? Yeah, so the way I think about these two categories is that observability is the system of record. That's where all the data is collected and put into context. So that's what we do as Instana is we take all the data metrics, logs, traces, profiles and put it into a system of record. By the way, in very high granularity, it's very important, so we do not sample, we have second granularity metrics. So very high quality data in that system of record where AI ops is the system of action. This is a system where it takes the data that we have, applies machine learning, statistical analytics, et cetera on it to figure out, for example, the root cause of problems, or even predict problems in the future and then suggests actions, right? What the next thing that AI does is, it suggests or automates an action that you need to do to, for example, scale up the system, scale down the system, scaling down because you want to save cost, for example. These are all things that are happening in a system of action, which is the AI ops space. Yeah, when I think about what you're talking about in terms of observability, I think, well, who needs it? Well, everybody is probably the answer to that. Can you give us maybe just a couple of examples of some clients that you've worked with in terms of particular needs that they had and then how you applied your observability platform to provide them with these kinds of solutions? Yeah, I remember a big e-commerce vendor in the US approaching us last October they were approaching the Black Friday, right? Where they sell a lot of goods and they had performance issues, but they only had issues with certain types of customers and with their existing APM solution, they couldn't figure out where the problem is because existing solutions sample, which means if you have a thousand customers, you only see one of them as an example because the other 999 are not in your sample. And so they used us because we don't sample. With us, if they have more than a billion requests today, you see every of the one billion requests and after a few days, they had all the problems figured out and that was one of the things that we really do differently is providing all the needed data, not sampling and then giving the context around the problem so that you can solve issues like performance issues on your e-commerce system easily. So they switched and you can imagine switching a system before Black Friday, you only do that if it's really needed. So they were really under pressure and so they switched their APM tool to Instana to be able to fulfill the big demand they have on these Black Friday days. All right, so before I let you go, you were just saying they had a high degree of confidence. How were you sweating that one out because that was not a small thing at all, I would assume. Yes, it's not a small thing. And to be honest, also it's very hard to predict the traffic on Black Fridays, right? And in this case, I remember our SRE team, they had almost 20 times the traffic of a normal day during that Black Friday and because we don't sample, we need to make sure that we can handle and process all these traces, but we did pretty well. So I have high confidence in our platform that we can really handle big amounts of data. We have one of the biggest companies in the world. The biggest companies in these worlds, they use our tool to monitor billions of requests. So I think we have proven that it works. Yeah, I'd say you're smiling too about it. So I think obviously it did work. It did work, but yeah, I'm sweating still. Yeah. Never let him see you sweat, Mirko. I think you're very good at that. And obviously very good at enterprise observability. It's an interesting concept, certainly putting it well into practice. And thanks for the time today to talk about it here as part of IPM Think to share your company's success story. Thank you, Mirko. Thanks for having me, John. All right, we're talking about enterprise observability here. IPM Think, the initiative continues here on theCUBE. I'm John Walls, and thank you for joining us.