KubeCon + CloudNativeCon EU 2025 | Alois Reitbauer, Dynatrace

Clips
More from KubeCon + CloudNativeCon EU 2025

Alois Reitbauer

Chief Technology Strategist

Dynatrace

play_circle_outline Unlocking Platform Engineering's Potential: The Interplay of Observability, AI, and Developer Velocity

play_circle_outline Agentic systems and AI as the next evolution of AI

play_circle_outline Real-time observability crucial for scaling AI initiatives and quick remediation

play_circle_outline Enhancing Digital Business Compliance through Real-Time Observability and Monitoring

Info
Transcript

Alois Reitbauer, Dynatrace

Alois Reitbauer

Chief Technology Strategist Dynatrace

In this episode of “Inside the Digital Business with Dynatrace,” recorded live at KubeCon + CloudNativeCon EU 2025 in London, we kick off the series with Dynatrace Chief Technology Strategist Alois Reitbauer as he talks with theCUBE Research’s Rob Strechay. As part of this ongoing exploration into how businesses are transforming amid digital complexity, Alois unpacks why observability is no longer optional — it’s foundational.

From the rise of agentic AI systems and platform engineering to real-time compliance and the evolving role of developers, this ... Read more

explore Keep Exploring

What is the importance of observability for developers and how does it enable velocity in organizations transforming to digital businesses? add

What is the next paradigm shift in software development that involves moving from static software to orchestrated agents with LLMs playing a role? add

What are the potential risks associated with AI systems and the importance of immediate detection of issues in real-time systems? add

What are some key considerations for businesses when it comes to compliance in the era of digital transformation and continuously evolving software systems? add

bolt Powered by CUBE AI

Alois Reitbauer, Dynatrace

search

Rob Strechay

>> Hello and welcome back to KubeCon CloudNativeCon London where it's actually sunny. Pretty amazing. It's like one of the first times I've been here and it's actually been sunny. We're also kicking off a new series where we're going to be going inside the digital business with Dynatrace. And I couldn't think of anybody better to help me kick this off than Alois Reitbauer who's the chief technology strategist. I got to get that off my tongue there for a second. But for Dynatrace-

Alois Reitbauer

>> Yes....

Rob Strechay

>> I think one of the things that I've been going around, I went to a lot of the day zero stuff yesterday here at KubeCon to hear what people were trying to understand and observability is having a moment. And I think AI's having a moment and AI and observability are having a moment where people are trying to simplify how they do things like platform engineering. And I think that there's really some neat stuff going on. And we just saw each other at Perform only a couple months back here.

Alois Reitbauer

>> It has been a couple months, it feels like weeks, but it's been months. Yeah.

Rob Strechay

>> I know it felt like yesterday, but it's really a key. And I think there's a lot of people here in particular, there's the role of developers in maintaining velocity and developer observability and how that really plays into what organizations are trying to do as they transform their businesses to be digital businesses. How do you see that really playing out and how do you see that velocity really being enabled by that?

Alois Reitbauer

>> I think it's key. When we started to talk about observability, it was very much an SRE-driven conversation, keeping systems up and running, ensuring that they work, know why they break. But when we talk about developers, they need more information. They're at the end of the day, the people who need to fix the application, and it's critical that they get the information to do so as quickly as possible. Even at Dynatrace, we measure this at this distinct time. How long does it take from when an incident occurs or we see any issues until the developer knows how to fix it? And the key is really giving developers the tools to do this and provide the analytics capabilities. And interestingly enough, observability for developers also, it's using the same data is different than observability for site reliability engineers. It's really understanding the very details and also visualizing the data differently. As an SRE, you might easily be responsible for a thousand applications and overseeing this massive fleet of applications. Very often when you're called in as a developer, you want to see that very single transaction that failed. You want to see all the information in context for this trace. What were these log statements that were called, you make a call to an external services, what was the response time? What was the load on this? So giving developers this context so that they can actually triage problems faster and also get a deeper understanding on how to manage and optimize their applications.

Rob Strechay

>> Yeah, and I think that is a perfect message for the folks here because I mean, we're surrounded by 14,000 developers or maybe not 14,000, but probably 10,000 developers here today. And I think one of the things that has been not only a big set of noise and big theme for KubeCon CloudNativeCon but in general is really around agentic systems. And they're having their moment as the next evolution of AI where it's combining things like LLMs and traditional AI or ML together to go and do an action or do a chain of thought type of thing. What should business leaders really understand about this shift and how should they really prepare themselves for that?

Alois Reitbauer

>> I think this is going to be one of the biggest paradigm shifts we have seen in a very long time. You obviously remember the shift we made from monoliths to microservices. Now we move to, again, from static software to agents that get orchestrated on the fly where LLMs play a role. Suddenly the dynamics and the execution of these environments become highly, we wouldn't say unpredictable, but way more depending on the individual interaction that's happening. And in all the conversations I'm having, I see a very interesting shift. Usually even with observability, we started to build microservice applications very early on. And then observability came in later, not calling it an afterthought, but it came in later. But I now see my conversations about talking about agentic AI. So really systems that do planning that reach out to other agents that change really behavior, sorry, based on the data that they receive and become very dynamic, like every interaction is different. I hear people say, "We are not going to put this into production unless we have observability." I think observability becomes very critical there. And also observability kind of gets even slightly redefined. So depending on which audience you're talking to. Sorry. It's been a lot of talking here.

Rob Strechay

>> Yeah, and it's dry in here.

Alois Reitbauer

>> It is dry here, yeah.

Rob Strechay

>> Which is again funny for London. That's putting it mildly.

Alois Reitbauer

>> It is perfect weather for London, unfortunately we're inside. Yeah. Now coming back to this, we have a number of target audiences. We have the very traditional data scientists, data engineers who build models, who fine-tune models, who optimize them. They care about model quality, model drift, optimization performance. Then we have the SRE. And for most of the SREs, GenAI workloads or large language models as small language models, they're just another service they have to run for performance. They have to look into capacity planning, they have to do all of this. And then we have what I call the AI-native developer, a developer who writes code applications that have AI as a central part of the execution flow. They use tools like LangChain, LangGraph, use one or multiple large language models, do the prompt engineering. In some cases they might even be responsible for some of the fine-tuning or creating some of the document on the rec pipeline side. And these people find themselves now in a situation where they have the most dynamic, most unpredictable applications they have ever had that they need to optimize. And then tying back to the first question, what does it mean for developers? So suddenly what they need to do, they need to have end-to-end tracing. They need to be able to debug in production like what we call live debugging, more or less, setting a non-breaking breakpoints. They need to see all the information in context for very specific transactions that the user's looking at that will never occur again. So it's putting a lot of this stress on developers as well and request new capabilities. And that obviously then last, but at least there are the line of business owners, I'm spending 4 million a year on a GenAI application. The main question I hear is not even ROI, like am I getting the money back? Even more often we're hearing our users are actually happy. So we start to measure user happiness and there's obviously many ways in how to do that.

Rob Strechay

>> Yeah, I think you hit on a lot of things there because I think there's just so much going on with AI. These systems have become that much more complicated. SREs lives weren't easy to begin with, but now these applications are growing in complexity. They're much more distributed in nature in some cases, and they really have just like a need that that real-time nature of these applications really is driving them. And in fact, kind of going back to that, why is the real-time observability for organizations just really required for them to scale their AI initiatives? Why is real-time observability powerful?

Alois Reitbauer

>> So real-time is really, you have to act as things are occur and you have to take counter action. So if you would look at it later on in life so a lot might've gone wrong. So once the first interaction goes wrong, and a lot can go wrong in AI, we have heard of situations where AI agents suddenly start to sell flights that do not exist or they leak PII data. In other cases when you need to find this out immediately, and as these are real-time systems, and as businesses move to more or less delicate parts of their business functions to AI agents, it basically means if the AI agent is not working, your business is no longer working and you won't have the people to actually compensate for what the AI agent is supposed to do. So having the information the first time something goes wrong, having the information transfer to the developer all they need, because you can't go back and say, "Run this transaction again," it actually might never occur again. And then being able to remediate very quickly whether you switch back to an older model version or you have to change some of the prompts very quickly or even change your guardrails so that some replies are not being sent back or reroute some of the traffic maybe to a different model. Unless you do this, your business is actually no longer working because your users... And I think it's also very interesting how we learned to work with technology. You for sure remember the times when we went to the airport two hours in advance, had everything printed out and looked at it last night while today we just take our phones out, get a reminder, "Oh, you should go to the airport right now," go there, just check in on our phone. Most that I bring is cabin luggage going through. And you know what happens when your phone's not working? It's all chaos. You haven't even planned for this anymore. And I think we will see this happening very soon in agentic AI systems that they take over very complex tasks for our users. We will get used to it very quickly. It won't take us years, it won't take us months, it will take us days if you have done it successfully five times and then your users will be piling up because they can't use the service anymore. And you as the business owner must be able to remediate super quickly. And if you would just learn hours later, you will have lost so much revenue that it would be very hard to compensate. And obviously user loyalty users would go somewhere else as well.

Rob Strechay

>> Yeah. Well, I think that's it is reputation, churn, things of that nature that business leaders really worry about and it's associated all the way down to the code and the reliability of the code. One other thing that you touched on and I think is really interesting is the role of observability for digital businesses, specifically around compliance and risk management. Because that to me would seem like a place where observability, you want it in first, not an afterthought to help you manage that.

Alois Reitbauer

>> I think that's where we see as more and more people like really now start to run their core business processes on software. We have been talking about digital transformation for a very long time. I think we are finally at the point where every business out there would agree that a substantial part of their businesses coming via digital in most industries. And compliance becomes key because you have so many compliance and that they have to follow. And compliance is also changing. So when we talk about compliance at Dynatrace, we talk about continuous compliance and also compliance in real time. Compliance historically was you took a snapshot of your environment and then you check it for compliance. And with modern software development, 10 minutes later you might have deployed something new and the whole check is not worth anything anymore. So even looking at compliance changes with observability data being available, you can switch to real time. So instead of scheduling regular checks that are very fixed when you run them, you will run the checks continuously and even use methods like anomaly detection, root cause analysis why you're no longer compliant. So lot of these best practices we had for running software systems now also start to make it into compliance.

Rob Strechay

>> Yeah, because seeing we are here in London, which used to be part of the EU, not in fact part of the EU right at the moment, but you have things like DORA and things of that nature. How do you see that regulation and stuff really tying into that continuous compliance that you talked about?

Alois Reitbauer

>> I think it's a key point. That's also why we started at Dynatrace to work on having what we call a compliance assistant. So being able to do compliance monitoring for our customers. And a key point is do it continuously because you want to know when you fail, because otherwise you end up in forensics.

Rob Strechay

>> Yes.

Alois Reitbauer

>> You don't want to be after the fact. You want to act in real time. If you see that something is not working properly, there might be some data leakage or something not working. Ideally, you can react very quickly, avoid fines, and even don't have to explain to anybody why it's not working. And I think that's also what regulators will start to realize more and more SSE observability systems actually being used for compliance. Oh, it's not like that one point in time. We check it every second week of March every year. But then we realize, okay, we want to get this every week or every time. We might just come in randomly, or we want it to proactively come back to us on incidents that you're seeing. And we also see that the reporting requirements become much stricter. If you see compliance violations with the 24-hour response time, we have to explain the entire issue to legal bodies, how it's working. I think that mindset shift will continue in this space as well. And obviously DORA is key. I mean, we run on digital money and it's a part of the critical infrastructure, and I think you'll spend across all areas of critical infrastructure where there will be way more liability and way higher standards on how to run these digital services.

Rob Strechay

>> Yeah, no, I totally agree. And I think we see that in spades. And again, I think that is probably a good place to leave this. I think, again, when you look at inside the digital business, it's really about how people are making those strides. And the people here, the 14,000 people around us are definitely out there trying to figure this out. And I agree with you, if you don't monitor it, you can't manage it and you can't really take action. So you have to observe it to begin with. But thanks for coming on board Alois. Really good.

Alois Reitbauer

>> Thanks for having me.

Rob Strechay

>> I'm glad to be kicking this off with you guys. This is a lot of fun and I think a critical thing to really explore. So thanks.

Alois Reitbauer

>> Thank you again.

Rob Strechay

>> And thank you for watching this episode live from KubeCon CloudNativeCon London, also inside the digital business with Dynatrace. Stay tuned. We got a little bit more to come.