KubeCon + CloudNativeCon EU 2026 | Michael Beemer & Jonathan Norris, Dynatrace

Clips
More from KubeCon + CloudNativeCon EU 2026

Michael Beemer

Principal Product Manager

Dynatrace

Jonathan Norris

Director, Software Engineering

Dynatrace

play_circle_outline Dynatrace Acquires DevCycle to Embed Feature Flags, Decoupling Releases from Rollouts and Slashing Deployment Risk

play_circle_outline AI/agentic development accelerates feature creation, increasing reliance on feature flagging

play_circle_outline Product managers and non-developers can own release timing and KPIs through flags

play_circle_outline Feature grouping: wrap multiple flags under a business-oriented "feature" for context

Info
Transcript

Michael Beemer & Jonathan Norris, Dynatrace

Michael Beemer

Principal Product Manager Dynatrace

Jonathan Norris

Director, Software Engineering Dynatrace

In this interview from KubeCon + CloudNativeCon EU in Amsterdam, Michael Beemer, principal product manager at Dynatrace, joins Jonathan Norris, director of engineering at Dynatrace, to talk with theCUBE's Rob Strechay and Paul Nashawaty about how the shift from big-bang code releases to progressive feature delivery is redefining software deployment in the cloud-native era. Norris explains that OpenFeature — which Dynatrace helped launch over four years ago — is reaching an inflection point comparable to where OpenTelemetry stood at a similar stage, with SDK d... Read more

explore Keep Exploring

How is Dynatrace evolving from an IT-ops/observability focus to support developers through DevCycle and feature flags (including feeding Dynatrace insights into the IDE), and what benefits do feature flags provide to developers and operations teams in large organizations? add

How will the rapid pace of AI development affect engineering teams' deployment cycles and release practices, and what role will feature flags play? add

How can feature flags change the release process, enable product managers and other teams to control feature rollouts, and help measure a feature's impact? add

How can feature flags be grouped into higher-level "features" and linked with performance and business telemetry so teams and executives can see what is being shipped and measure the impact? add

bolt Powered by CUBE AI

Michael Beemer & Jonathan Norris, Dynatrace

search

Rob Strechay

>> Hello and welcome back to KubeCon + CloudNativeCon EU 2026. We're here from slightly cloudy Amsterdam today, not too shocking. We're into spring and allergies are kicking.

Paul Nashawaty

>> Oh, they definitely are.

Rob Strechay

>> To put it mildly. Paul, joining me here again and we're going to be talking about one of our favorite things here. We're going to be talking about feature flagging and OpenFeature with the folks from Dynatrace. So welcome on, Michael Beemer, who's principal product manager for Dynatrace. Welcome on board.

Michael Beemer

>> Yeah, thanks for having me.

Rob Strechay

>> And we have Jonathan Norris, Director of Engineering for Dynatrace as well. Welcome on board.

Jonathan Norris

>> Yeah, thanks for having me.

Rob Strechay

>> I mean, to me, this is, again, one of these favorite things for me to talk about because as having been and built product, going from using a spreadsheet to actually having a way of approaching and flagging features to bring them out, test them or bring out certain things to different customers is super important. So I'm really excited. Kind of help us understand what's been going on this week with OpenFeature. And now that DevCycle is part of Dynatrace and how things are going.

Jonathan Norris

>> Anyway, yeah, it's been a really great week overall. I think this is the biggest collection of OpenFeature maintainers and contributors we've had at a KubeCon. And so we spend a lot of time just sitting down together face to face, hashing things out. A lot of those maybe difficult discussions you have in open source. You can really get through a lot of the details when you're sort of sitting face to face. So it's been great to really see the progress and how OpenFeature is growing. I think we're at that phase where we're starting to see that exponential growth in the downloads of OpenFeature SDKs and we're seeing large companies like Google get involved as contributors and Datadog. And, obviously, Dynatrace has been the main sort of contributor to OpenFeature for a while getting it started over four years ago. And it's just getting to that phase where you can see like it's getting exponential. And I think it's where maybe OTel was like four years ago is kind of where I think OpenFeature is starting to get to. And I think more and more large platforms and companies are using it. And we always sit at the booth and every time we find a new company that's using OpenFeature that we didn't know of before, and that's always like a fun experience for all of us.

Paul Nashawaty

>> Definitely, definitely. Well, I'll say that coming out of Dynatrace Perform, Dynatrace historically has been known to be like the focus on the IT ops, the focus on the observability space. Now, you're pivoting to the developer. DevCycle is really kind of driving it. The feature flag pieces are really, really important to the developer. What I really like of what you're doing is you're feeding the insights from the traditional Dynatrace platform into the ID, directly into the ID. So can you talk a little bit about that? No, I won't say a pivot, but like the adding on to the ITOps and the developer persona.

Michael Beemer

>> Yeah, I think it's kind of covering both roles. So you have to have a great developer experience, because developers are the ones that kind of add feature flags to your code so that feature flags start and end with developers, but really like the operations team need to be on board as well. And I think they can help sell that to maybe the first development team. And then like a large enterprise that starts there, they're proven to be successful and that can spread throughout the organization.

Jonathan Norris

>> And the core value for feature flagging for a big organization is when you can separate releasing code from releasing features to your customers. And so when you make that logical separation, it really unlocks your teams to ship to production more efficiently, more quickly, and really derisk those like big bang code releases to production. And so whenever I'm telling people about feature flags, I'm like, "Imagine a world where you didn't have to worry about the code releases, because you weren't actually releasing new functionality or code changes to your end customers until you rolled them out progressively through a feature flag."

Rob Strechay

>> Yeah. I mean, I think that was part of it when we've talked before. I mean, getting to trunk, right? How do you not have to go through all these different merges and things of that nature? Is that been a lot of the, like when you're talking to these folks, again, having less different streams that you have to bring together by having feature flags and then putting it in and then bringing it through? Has that been part of the ?

Michael Beemer

>> Yeah, absolutely. Basically, just trying to get features merged into main as quickly as possible. It doesn't mean it's released to customers. You can keep shipping away at it. You can have product managers take a look at it in production before customers are impacted and it kind of just changes how you release software.

Jonathan Norris

>> And with the pace of like AI development, just making, generating and creating new features so much faster, I think most engineering teams are really going to start to feel the pain of their slow deployment cycles, right? And so you look at some of the biggest AI companies out there, the OpenAIs, the Anthropics, people like that, they're shipping a feature a day and all those features are behind feature flags. So they're testing internally in production, using feature flags as a core part of all their workflows. And so if anything, in the new wave of agentic development, like feature flags are going to be a core release and delivery process to do that. And I think it's only going to get more and more important. And that's why even standards like OpenFeature, become more and more important because now the agentic systems understand, hey, OpenFeatures, the standard like OTel, if you say, "Hey, I want to add feature flags to my app," the AI is probably going to recommend this open standard and the extensibility that comes with it.

Paul Nashawaty

>> Yeah. Jonathan and both Michael, you both mentioned this, that the acceleration of the SDLC, the acceleration of the development delivery of applications, we're seeing in our research, in 2025's research, 24% of response to global survey, 24% of respondents indicate that they want to ship code on an hourly basis, yet only 8% are able to do so. Adding feature flags into the CICD pipeline really helps with allowing for that, what's a P0, what's a P1? How do you push it out the door? Is that where you're seeing the kind of intersection between the CICD pipeline and the ITOps side?

Michael Beemer

>> Yeah, absolutely. I mean, you're able to ship a lot faster, but that's also where maybe the risk comes in or the opportunity for observability tools. If you're kind of blind to the impact of a feature flag, you're introducing new challenges then. And so I think that's where it's an ideal pairing, bringing the DevCycle team and building a new offering in Dynatrace and kind of combining the worlds of like faster releases with the safety and oversight that a observability tool brings in.

Rob Strechay

>> Yeah. I guess that's where I was going to ask, because like when you traditionally APM and feature flags don't ... You don't usually think of them together, but to me it makes sense, "Hey, I put these five new features in, I have them flagged on or off, I know which customers are doing that." So you're limiting your blast radius. Has that been one of the things that you've seen from customers as well?

Jonathan Norris

>> Yeah, like most customers when releasing large features, if I just roll them out to a small percentage of users or like they may have a ... One of the features we have within DevCycle is you can set up beta test groups where people can actually opt in to being like, "I want to opt into the beta of this feature." It might be fully in development feature that somebody is shipping and those enable you to ... We can then connect that with the observability to data and say, "Okay, we see this new error report that's causing, I don't know, an issue or a crash in our backend service, but we can know that it's tied to this feature flag that's only enabled for like a couple customers." A couple people are seeing it and having that insights allow you to be like, "oh, okay, maybe we need to roll that back, make some code changes and move forward with it." And I think you can see where things are going once you tie observability, feature flags, AI together, when you connect all those things together, there's a lot of power that can come from it.

Paul Nashawaty

>> Actually, I love where you went with that because I think that the fact is that the two barriers that I see organizations talking about with overcoming these challenges to deliver applications is complexity and skill gap issues, right? Those are two areas you hear all the time. And when you're looking at the code to deployment to runtime validation, there's so many pipelines, right? What Rob was saying, like you take the observability pieces in, that observability pipeline, the APM pipeline, and then you have to tie that into that DevSecOps pipeline, right? And those actionable insights, we're seeing everything shifting left or shifting everywhere really. And you see that shift happening, but there's a lot put on the developers. How is this helping with reducing complexity and skill gap issues?

Jonathan Norris

>> Yeah. I think one of the things it can do is it can actually enable other parts of the org to own sort of that release process, right? And so product managers can now be the ones who make maybe the final call of like, is this feature ready to go or not? If you're not tying a big bang code release to production that like, I don't know, you have 10 teams trying to ship one code base to production and trying to line up all these different things to do this big public announcement, that can be really risky. I think every manager of every big any tech company has had issues with that in the past. And by chunking things out into feature flags, you can say, "I'm going to do a release of this feature today because it's ready. I'm going to do this other release of this feature later in next week." And all that code might actually have been in the same code release a week ago, but you can sort of plan out those releases and enable those features and product managers and product marketing and other parts of the organization can actually help own and control the timing of those and be involved in that.

Michael Beemer

>> Yeah. And like one of the concepts we're going to include in our offering is kind of wrapping multiple feature flags together under a feature. And so like the feature's meant to be like encapsulate the business context around the release. And we think that could be quite interesting too. If you could bring in like performance and like business telemetry there, you could actually see what the impact of a feature was where a lot of times it's like PMs may have good intentions around collecting KPIs on a particular feature, but if we could do that automatically, then it's something you could just get in the habit of tracking.

Jonathan Norris

>> This has been one of the core features of the DevCycle platform where when you look at our dashboard, you look at the DevCycle platform, you're not looking at a long list of feature flags that some developer made that you have no context is. You're looking at, "Oh, I'm shipping the new checkout flow, the new onboarding flow, the new whatever feature." And you're seeing at a high level, these are the features my teams are shipping and they're assigned to these specific teams. So it's almost like as a CTO, as a higher level executive, you can look at our dashboard and be like, "Here's actually what's in flight. Here's what actually is shipping to prod." And then within those features, you might have one to, you might have I've seen hundreds of flags under a feature before.

Rob Strechay

>> Yeah. I mean, I think when you look at it, I mean, again, going back to my days of actually shipping code, right? And to your point about who can then go and ship it, it becomes a lot more fun for being able to see that business context and relate it back. Because for instance, it may take multiple sets of sprints to actually finish out a feature or maybe you get through three out of five and you're like, "Hey, it's good enough to go for what we want to test out." Are you seeing as from an OpenFeature perspective that, to Paul's earlier discussion, the personas are almost changing and people who are coming in and looking and trying to figure it out, like do you get product managers and others outside the development community that are coming in and looking at that?

Michael Beemer

>> Yeah. I mean, for OpenFeature specific, I think it's probably the most attractive to like engineering teams. Feature flagging concepts supply across a lot of personas though. So I think here at KubeCon, it's a little interesting. There tends to be more operation people, so we try to hit it more for like the safe releases and thinking it from that angle and then hopefully then they go back and tell their development teams to take a look at OpenFeature.

Rob Strechay

>> Yeah. I mean, I would assume from a platform engineering team, they're sitting there going, "Hey, this can help us help them go faster." Is that kind of one of the things?

Michael Beemer

>> So we've seen that too. Yeah. So like at the last KubeCon actually, we had a guy come up to us and said, he was on the platform engineering team and he's like, "We don't really care what backend you use for managing feature flags, but they're recommending OpenFeature for all their teams." So kind of .

Jonathan Norris

>> I think for platform engineers, a lot of times they own that big bank code release and they have very little context to like what's actually in those code releases and they're responsible for getting that shipped to prod. So for a lot of them, like their teams adopting feature flags dramatically reduces the risk of those big bang code releases going wrong, right? So I think actually for a lot of those platform engineers, they should be evangelizing feature flagging and these methods to their teams because it actually shifts the responsibility to the teams actually building the features, right? If that feature's not enabled in production until the team chipping it has tested in production and the product manager or the QA folks have QA'd it in production, that really reduces the risk for those operational people and probably makes their lives a lot easier in the long term.

Paul Nashawaty

>> Yeah. So with that, I mean on that same vein, like there's an understanding of people that are like, "Hey, I use Dynatrace, I'm familiar with Dynatrace, I'm familiar with observability, I'm familiar with APM." I guess it's a good opportunity to say like, how easy is it to just try DevCycle? How easy is it to kind of bring this on board?

Jonathan Norris

>> Can I take that one then?

Michael Beemer

>> Yeah, sure.

Jonathan Norris

>> The nice thing about OpenFeature is that it's similar to OTel. We modeled the program after how OTel is set up where there's a standard set of SDK, the standard set of APIs that are consistent across languages, across platforms that you can implement, and then you can plug and play providers in there, right? We hear this a lot that there's a lot of companies out there that have maybe multiple feature flag providers or internal solutions that they've built up over the years. And so with OpenFeature, you can actually create multiple providers and sort of plug it all into one common interface and start the process of like, let's maybe over time slowly migrate those old internal solutions or to an open source solution or to a vendor like DevCycle. And so, yeah, the OpenFeature sort of platform in SDK gives you that ability to mix and match and combine different vendors and providers under a single common interface.

Rob Strechay

>> So this has been great, but I want to know between now and Salt Lake, what should we expect from Dynatrace feature management and OpenFeature?

Jonathan Norris

>> There's a lot of hard work going on behind the scenes, I can tell you that.

Rob Strechay

>> Yeah. Don't put them on the spot.

Jonathan Norris

>> Yeah. No commitments

Rob Strechay

>> Roll out your roadmap.

Jonathan Norris

>> Yeah. No commitments on the roadmap just yet, but from the OpenFeature side, I think we're really pushing towards how do we get this to a graduated project. With the number of large enterprises that are using it in production already, the big contributors, like some big names coming on board, like the Googles and the Datadogs and some of the existing sort of large enterprises that are using it. It's like, I think we're at the state where like, "Okay, this is almost like the API level layer is kind of settled. We're getting a lot of those things to 1.0. How do we become a graduated project and sort of grow this thing to the scale of an open telemetry as part of the CNCF?"

Rob Strechay

>> Awesome. Well, I really appreciate you guys coming on here and helping us kick off day three here. Been fantastic. Always good talking and a thing that's near and dear to my heart as well. So I mean, I just know the pain of what I used to do in a spreadsheet before there was things like OpenFeature, so thanks for coming on board.

Jonathan Norris

>> Yeah.

Michael Beemer

>> Thanks guys.

Rob Strechay

>> And thank you, Paul.

Paul Nashawaty

>> Of course, of course. Great being here.

Rob Strechay

>> Absolutely. And thank you for watching this episode of theCUBE, live from KubeCon + CloudNativeCon EU 2026, live in Amsterdam. We'll be back with more. Stay tuned.