Dynatrace is a company that really reinvented itself as it saw the world it played in changing over the last few years. They moved from tracing and APM to what the company calls software intelligence and smart monitoring.
In this DevOps Chat we sat down with Alois Reitbauer, chief technology strategist at Dynatrace, to discuss the company’s evolution from distributed tracing to its current focus on software intelligence and smart monitoring.
As usual, the streaming audio is immediately below, followed by the transcript of our conversation. Enjoy!
Alan Shimel: Hello, everyone, it’s Alan Shimel for DevOps.com, Container Journal, Security Boulevard, and you’re listening to another DevOps Chat. Our guest on this DevOps Chat is Alois Reitbauer of Dynatrace. Alois, hopefully, I got your name almost right, at least.
Alois Reitbauer: That’s very good. [Laughter]
Shimel: Okay. [Laughter] Alois, thanks for joining us. As I mentioned, you’re from Dynatrace, but why don’t you share with our audience kinda what your role is at Dynatrace right now?
Reitbauer: Yeah, so, my current role is, I’m working as the Chief Technology Strategist at Dynatrace, and in this role, I’m responsible for most of our cloud initiatives, our open source initiatives, and help to drive forward what we in the industry understand as APM observability, cloud automation, and so forth.
Shimel: Excellent. And, you know, Alois, I always like to ask people, because we have so many people in the audience who are at various stages of their careers, right? Give us, if you can, a sense of how you came to be where you are today.
Reitbauer: Yeah, that’s a good question. So, I think I started in the very traditional way. So, I did my studies in software engineering at the very beginning, and even during my studies, I was already working at a company, then took the usual journey of finding a first job once you’re done with your university degree, so you move away to a big city. And then I was working on some projects, but then realized that the project business was not so much what I liked. Then I ended up at Dynatrace, starting to work in the monitoring space again on different projects, and I have been doing this for almost 14 years, which sounds like a very long time—
Reitbauer: – but—
Shimel: It does.
Reitbauer: [Cross talk]
Shimel: But I know it goes quick.
Reitbauer: It does go quick, and the market changes so quickly. So, people sometimes ask me, “Aren’t you bored with what you’re doing?” And my answer, actually, is—no, I’m not. Because, at least every two years, something fundamentally changed in the way we’re doing things, and it’s just an entirely different job altogether again. That’s, I think, what is keeping me going here and what keeps the space being very interesting.
Shimel: I agree with you. I mean, just to give you a little insight into my—I mean, Alois, I went to law school and practiced law for a little bit. And then I got into computers just prior to the Internet going commercial, and started my first company, it was more in hosting and then infrastructure and I did ASB, then I got into info sec, we didn’t call it cyber security then, and spent about 15, 18 years in security when I kinda got interested in DevOps.
I was a founder, a co-founder of several venture backed companies and then started MediaOps, the company behind DevOps.com and Security Boulevard and Container Journal. So, I mean, that, I think, is one of the—and especially for our audience listening out here, don’t get pigeonholed into being a Java programmer or a program manager in mobile app development or something. You don’t know what the technology is going to be and where your career’s gonna take you, right, going forward and it’s important.
Reitbauer: Yeah, and I think all the technology changes a lot. Like, talking, for example, about the monitoring space. If you look at the last, like, 13 years, we started to do a lot on the service side with Meteor applications, and a lot moved more into the browser as we started to have sync to page applications, Java script heavy applications, and everything moved into—the focus shifted stronger into the browser where we talked about real user management and working a lot of, in this space. And then, now, we moved into, we moved back to the server side and all the microservices emerged. We had way more complexity on the server side again, and now we’re moving more into automation, which brings us to infrastructure topics.
And, at the same time, we have this whole change in the industry, really moving from also, in our case, enterprise software working more in open source type of environments. So, there’s really constant change on multiple levels, and I think the most important thing, really, is to be curious and to start learning, taking all of these opportunities that just opened up to you.
Shimel: I agree 100 percent. You know what, we went down a little bit of a rabbit hole, Alois. [Laughter] Let’s bring it back to Dynatrace, if we can.
So, I think a lot of people have heard the name Dynatrace and, you know, there are some people who may think we still just kind of associate Dynatrace with, like, mainframe computers or something. But, you know, this is a very different company, Dynatrace, than, let’s say, 5, 6, or 10 years ago or what have you. Give our audience, share with our audience a little bit, what and who is Dynatrace about?
Reitbauer: Yeah, so, for those who don’t know Dynatrace or have never heard of Dynatrace as well, we have been around for quite a long time. So, we were really very early on in the whole, what is now distributed tracing and ad observability space. In fact, Dynatrace was the first company that had a distributed tracing product to the extent that it really worked in production environments. So, everything we talk about right now in the microservice observability space, we were really the first ones to build this for a production ready environment 13 years ago.
And then, obviously, the market evolved, our product evolved over time, supporting more and more technologies. I want to go into all of the details of that stage, but we were like the one company that could monitor big environments that had broad technology coverage. And, back then—sorry?
Shimel: No, no—I said yes, I’m listening. I’m sorry to interrupt.
Reitbauer: And back in the days, the biggest challenge was, can we collect all the data that’s out there, and can we get all the data that we’re interested in? It was the big challenge in the very beginning where we were focusing most of our efforts.
But, like, five, six years ago, we realized that with everything that’s happening right now with microservices, cloud computing and so forth, the requirements drastically changed. So, we had all the data that we wanted. We had, actually, more data than we could even process, so, that’s where we started to move back then with this whole area of starting what we called answers instead of data, building more and more analytics on top of that data, and started to use the term artificial intelligence as we started to use these technologies in our next generation product back then, and then we were bootstrapping internally at Dynatrace.
Interestingly, this was before everybody was using the term AIOps, when we were doing this. And we built our own analytics layer that does all the data analysis fully automatically, so that was where we reinvented more of the company, what we were delivering, as a product.
And now, more and more, like in the next evolution we see, well, just having a monitoring product by itself or just for a monitoring team is no longer what you want. You want the monitoring product that’s now smart and figure out where things go wrong and how they can also potentially be solved to directly tie into the platform, which gets even more and more into a platform type of approach where I sometimes see or start to see use cases where people are not using the user interface so much, but really use the APIs to automate based on the outcomes of their monitoring data.
That’s really the journey that we’re on, and that’s also why we started to no longer use the term APM for what we’re doing, but rather focus on the term software intelligence. It’s really having intelligence and understanding of how a software system works and, in case it doesn’t work properly, how to get it back to a state where everything is fine.
Shimel: I agree with you. I mean, one of the things, I mean, we—another term I’ve seen thrown around is AIOps now, right? AIOps is kinda replacing what we used to call APM. And I’m not a big fan of the whole AI thing, to tell you the truth, Alois, because I think it’s a loaded term. But it’s not—I mean, it does kind of describe what you’re talking about, right? You know, intelligent monitoring and so forth.
Reitbauer: I think AI is also used for a lot of stuff, and there’s also the saying that if it’s written in Python, it’s machine learning; if it’s built in PowerPoint, it’s AI. And I might agree that there’s maybe some truth to it. And the term is overloaded, but at the very basis, when we built what we built right now, which is called the Dynatrace AI Engine, whether you wanna call it that way or not doesn’t really matter that much.
But the core piece for us was, we built—we have, as operations professionals or as SREs, we have a well-defined methodology, we have a well-defined understanding of how the system works. And a major part of our work is to analyze data and interpret it following certain patterns, and teaching, now, another piece of software to do the same thing that we used to do, being able to understand and perceive an environment to how things fit together based on whether it’s a rule based system, it’s machine learning, it’s some statistics, or whatever algorithms you use, and help us to do the things faster than we can do them or do them with environments—like, more environments that are significantly better, I think that’s where the real value comes in.
And whatever we want to call it at the end of the day, but I think it’s not just a nice to have today, it’s really a necessity, simply because we all don’t have the time to look into all of that data.
Shimel: Yep. I agree 100 percent. You know, we were—just, in terms of our audience, we originally were gonna talk over in Barcelona at the KubeCon event, and it was just crazy there, and we wound up saying, “We’ll do this podcast afterwards.”
I’m interested in, how is the whole cloud native Kubernetes kind of—I don’t wanna call it a movement, but certainly the migration to that sort of infrastructure, how is that affecting what you guys are doing at Dynatrace?
Reitbauer: I think it definitely provides a challenge, a very positive one, and at the same time, it also provides a great opportunity.
So, what do I mean by challenge? I think the challenge is that these environments are just way bigger and way more dynamic. In the old days—so, like, the early 2000s—most people were running suite tier type of applications.
Reitbauer: You had a [crosstalk] server, an application server, and a database. And you might say, “Well, I don’t really need tracing, it’s either the one or the other that’s broken,” and my environment is not that super complex. I would say that was also half the truth, because enterprises were running significantly larger infrastructures, but you kind of understood what was going on. And also, we didn’t have capabilities like we have today, and in Kubernetes, with self-healing, where things will start themselves automatically, scale up and down. So, we kinda knew what we were running.
Reitbauer: But this is not so much the case any more today, so we want our software to start to repair and heal itself, and environments are getting bigger. So, that whole movement from monolith to microservices, as great as it is to have more and smaller components that you can work with independently, it means that you have more components that you’re working with independently, which means more complexity and more entropy in your system, right?
So, we need to have this support, and for us, it’s really—how can we help a human to better understand what’s going on in the system and remove all the noise and really get them the information that they need to work with? So, that’s really the challenge, and that’s back to our AI discussion from before, where these type of approaches can really help to tell somebody, “Okay, this is what you really have to focus on, and this is what the actual problem is, and the other 400 things are just effects and not the cause.”
So, just a clear separation of cause and effect out of a myriad of data is a massive value that a software intelligence solution can provide. So, that’s more the challenge side.
The other opportunities, we cannot change systems more or less in real time, because everything has an API. If you look at Kubernetes, you have the Kubernetes API or the ________, and it can change the system at one time. So, if the monitoring system detected something is wrong, we can actually take action. Back in the days where it was manually deploying an application to a server, I could not take an action in real time.
Today, if I detect a problem—for example, in Dynatrace—I can kick off a script that remediates a change, that scales something up, that rolls back a deployment, that reconfigures, maybe, a service match to route traffic somewhere else or do all different kind of things, and it can really automate it. But back in the days, it could show it to a human and they then have to figure out how to do this, and maybe after the change request was going through and somebody was manually logging into the server, that wasn’t really possible.
So, while we’re confronted with more complex environments that we have to make sense of, the advantage is, if we actually make sense of them, we can take more immediate and real time action on the findings that we have.
Shimel: Sure. You know, this brings up, Alois, the whole thing—so, I’m a little older, right? [Laughter] And I often tell some of the younger folks I deal with in DevOps and stuff is—you know, just be thankful for what a great time to be alive this is, to be in this ________ world with stuff like this, compared, you know, you’re doing this, what’d you say, 14 or 17 years, right?
Reitbauer: At Dynatrace—the whole thing, a bit longer, so, I’m also not that young any more.
Shimel: Yeah. I mean, think about what it was like 20 years ago, right, where an Ops guy setting up instances or a service maybe had a little tool belt with screwdrivers and pliers, right? [Laughter] Because you were literally rack mounting, and stuff like that.
So, it is—I mean, we live in such interesting times, and there’s so much, so many options, right, of how you can do things and what you can do. I think sometimes we forget or we take for granted that, you know, what the cloud has brought, and what the Internet itself has even brought, let alone things like Kubernetes and cloud native and some of the stuff we deal with today. So, I just—you know, I’m constantly amazed at it, and I don’t think we should take that for granted, either.
Bringing this back to Dynatrace, though, with the emphasis on open source and cloud native, does it make it harder for companies like Dynatrace that have some great legacy customers as well, right? I mean, you’re not gonna throw out the baby with the bath water, right? You’ve gotta serve—how do you kinda, not serve two masters, because your old customers are doing this fresh stuff, too, am I right?
Reitbauer: I think everybody’s doing it. So, every company is at different maturity levels, obviously, and if you have been around, maybe, since the mid-‘70s and your technology stack is based on mainframe and other things, you’re just not throwing everything out and making everything cloud native.
But everybody wants to go there, and I think everybody is on that journey and needs support on it. Obviously, if you start over today, your environment is going to look different than if you have been doing this for the last 40 plus years, so.
Shimel: Yeah, absolutely, I agree. Agreed. Alois, we’re coming up on the end of our time—I told you this goes quick, I apologize. But anything else happening with Dynatrace, maybe, you wanna share with the audience, or?
Reitbauer: Yeah, so, you mentioned open source, which I think is also interesting, because most people maybe think of Dynatrace as an enterprise software company that provides enterprise software, which we obviously do, and how we’re making our money, you brought up in the last question as well, does anything change for us, and I think the change is that we are way more engaged in open source, where we see open source as a great opportunity.
So, we, for example, are part of the OpenTelemetry project that a lot of people maybe don’t know about us, where we actively contribute to this project building and more and more observability tech, especially in the cloud native space. Also, just, not that recently, but over the last half year, we built our own—we took our own best practice that we had and we invented Dynatrace and moved to what we call, like, the NoOps approach. So, no manual processes and full automation, and only get humans involved to solve the problems that automation cannot solve.
We took this entire stack and started to put it into an open source project which you call Keptn, that we now also provide to the broader community and share. Because we, as you mentioned, we’re living in a very interesting time right now, and I think it’s really about sharing all of our expertise with as many people out there, and also learning from each other.
So, some things we’re doing today—just think about blue green deployments. If you really go back to e-competency, how long have you been doing this? What’s your experience? Do you usually have people who have 10 years’ experience of doing blue green deployments in cloud native? Obviously not, because these people do not exist, or there are some, very rare, from the very early start of it.
And, it might feel counterintuitive for a company to sell software to do all of this. I think you have to be part of that community and actively work there, and there is always a place for a commercial company to work with, and you want to work with somebody who can both support you on the hardware enterprise, relay the problems, but also give you the security of the commercial software company, while also being sure that these are also the people who bring the industry forward and move to the next level. And I’m also a strong believer that as we work on these open source projects and support the industry to be more mature, there will become more interesting problems. There will always be something for a company that we can engage with and find a good way to provide value to our customers.
Shimel: Agreed, agreed. And again, this is another way the industry’s so much different than it was 15, 20 years ago. Sounds good.
Alois, we’re about out of time, here. I told you it really does go quick. For people wanting more information—obviously, Dynatrace.com.
Shimel: Any other kind of resource you want to mention?
Reitbauer: Yeah, obviously, just Dynatrace.com is great for the website. I also want to mention our PurePerformance podcast that’s done by a colleague of mine, Andy Grabner. They have a lot of information related to Dynatrace, but not just Dynatrace—really, across the entire industry, and so, Andy’s always interviewing, whether it’s one of our customers or other companies out there, people in the industry. Great resource. If you want to learn more about what we’re doing on the open source side, I really recommend having a look at Keptn, the Keptn project, which is Keptn.sh, yes, you can buy .sh domains.
Reitbauer: That’s definitely a great resource, and obviously, also, our Twitter channel where we post a lot of the work that, what we’re doing, and that’s also an easy way to reach out to us.
Shimel: Excellent. Alrighty. Hey, thanks for being our guest on this episode of DevOps Chat. Alois Reitbauer from—I still got it right?
Reitbauer: Yes. Perfect.
Shimel: From Dynatrace—hey, we’ll check back with you in a few months and find out what’s happening.
Reitbauer: Yeah. Thanks for having me.
Shimel: My pleasure. Alright, this is Alan Shimel for DevOps.com, Container Journal and Security Boulevard. You’ve just listened to another DevOps Chat.