PurePerformance - Safety First in Multi-Cloud Environments Alois Mayr

Episode Date: January 30, 2019

Alois Mayr shares a few new ideas about Dynatrace and Kubernetes, in addition to looking into what Dynatrace plans to release this year.

Coming to you from Dynatrace Perform in Las Vegas, it's Pure Performance! Hello from Dynatrace Perform 2019 in Las Vegas. I'm Andy Grabner and this is Up Close and Personal with product management on Pure Performance. I want to introduce my guest, Arliss Meyer. Hey Arliss, I just bumped into you. I think you just came out of your session. Yes, hi Andy. It's been a while since I've seen you like last time. For people that don't know, we work in the same, well not in the same office, but at least in the same location in Linz, separated by the Danube, right? Yeah, separated by the Danube, but right now we are both in Vegas and having fun here. Hopefully the background noise is not too bad for the listeners. So Alois, you just came out of your breakouts and I want to ask you a couple of questions, especially for people that were not able to join before and couldn't attend your breakout.

So let's get started. What was your breakout about? What was the title and who was presenting and what was the topic? Yeah, so the breakout session right now, it was really just a cool story from Copa Airlines, which is like the largest airline in Panama and Central America. And the session was about safety first. So because it is airlines, safety is always important, right? And they actually run a multi-cloud environment. So they have like tons of stuff running on AWS, on Google, Azure. They also have stuff running on prem, VMware based clusters. And it was great to see what they actually did

and how they used Dynatrace to make sure that their systems are working properly. So now, multi-cloud, typically when I hear multi-cloud, I assume they run the same system on different clouds, just as a failover, high availability or maybe cost so that they can switch over for the least expensive, the look that is close to the end user. Is this the same thing for Copper or is there a different strategy?

No, no. So they follow a different strategy. So what they basically did was they used the cloud that best fits the case, what they want to solve. So they use, for instance, AWS for running their website or also like the web check-in. So whenever customers of the airline want to check in, they will use basically AWS offerings. But on the other hand, they have stuff running on Azure

for API management because in Azure, there is like a decent service around API management and that's why they want to use that. And on the other hand they have Google Cloud where they use business intelligence services from Google because Google is well known for having like proper means to analyze data and all that stuff. And then you also said they have something on-premise as well, so what's still on-premise do you know?

Yeah, so the most important, let's say, applications they have, they call them required to flight. So this is like a group which is like an airline terminology, it seems like, where they run crew management and baggage systems. So like all these applications, they run still on-prem on their VMware cluster. And so first of all, thanks for that explanation, because I always kind of assume you just run the same thing on multiple clouds and then switch, but this is a different idea. Yes.

Now, do they have these applications that run on different clouds also talk with each other? Of course, of course they do because you know if you run API management on Azure, of course what we would like to manage I mean, it's the APIs, but they need to communicate with stuff on AWS and on other platforms, so of course they have like connections between this cloud environments. Alois, you are a technical product manager here at Dynatrace, responsible for cloud technology. Maybe just give us a little background on what your field is. Yes, I'm mainly working on Kubernetes and container-related topics. And I have a bit of history in other cloud technologies. And that's what I'm mainly working on. So bringing like full-stack monitoring to whatever is running based on Kubernetes, Cloud Foundry and Cloud Technologies.

And interestingly, Copa Airlines also is using Kubernetes already, so it's a very modern airline. And they use Kubernetes mainly for DevOps related workloads. So they build pipelines and the deployment automation. That's pretty cool. So now from a product management perspective, what else is there that maybe you and your team, I mean I know that you as a product manager, but there's a larger development team behind you. What are some of the capabilities that maybe Copa Airlines has influenced by you talking with them? Or what did you and your team bring to the table that Copa Airlines is benefiting from and maybe other customers too? So what are some of the core things?

Well, I think the most important piece here is to have a proper detection and awareness of the cloud technology within our product. Whenever you deploy the one agent, we will automatically discover the cloud the agent is running in and the application is deployed to. This is important because this allows us to properly detect all this meta information and how things are deployed and where they are deployed and this helps us to draw like a dependency map of everything that is running in different clouds and this also helps us to properly you know show and visualize the information to the audience in the Dynatrace UI so

whatever your role is in the customer's company, you can then select, okay, I'm only interested in this AWS stuff, or I'm a Kubernetes guy, I do not care much about what's running on AWS, or somewhere, I'm only interested in Kubernetes, and this is really helpful. So this is, I think, critical

to understand the context of the applications and the platforms they are running in. And on the other hand, of course, what we have also learned is it's always key to have the easiest way to roll out monitoring. Because if you want to have monitoring for applications, you do not really want to somehow bake in monitoring agents to images or something. It's really key to get it just out of the box because you want to focus on building the stuff you would like to build or running the stuff you would like to run and not taking care of how to best enable stuff to make it monitor able so that's I actually am doing a break up with SAP who is actually talking about automating

dinosaurs rollouts at scale right baking the monitoring the one agent deployments into that work terraform scripts okay you know using Bosch for Cloud Foundry using chef puppet answer but whatever and that's exactly what he said right it form scripts, using Bosch for Cloud Foundry, using Chef, Puppet, Ansible, whatever. And that's exactly what you said, right? It just has to be simple and easy, part of your deployment automation, and you should not worry as an application developer, for instance, how to get monitoring visibility. And I think the other thing you said, and people may not be

aware of this, having all this metadata about knowing which cloud it is, which host, I think we are pulling the metadata from Kubernetes, from Google, from Cloud Foundry, like in which namespace does it is, which host, or I think we are pulling the metadata from Kubernetes, from Google, from Cloud Foundry, like in which namespace does it run, right? In which availability zone does it run? This data is all available automatically, and we can use it for tags, and then tags can be used to management zones for the use case you mentioned, the filtering and also access control, right? Yeah, correct. Correct. And it also touches alerting, right? So you can even define, so I'm the Kubernetes guy, I do only care about alerts that are

related to my Kubernetes cluster, right? So you can even use it for that, which is great. Or if you go a level higher, if you're responsible for an app, you want to be alerted if the app has a problem and independent on whether it's a piece that runs in Google, in Azure, in AWS, or maybe on-premise. And that's the nice thing. And for customers that run many different environments, platforms, and clouds, this is very important.

So this not only holds true for CodePay Airlines Airlines, we do have other customers as well that run like stuff from Kubernetes and Cloud Foundry together in the same data center sometimes in different clouds. And this is perfectly fine. They really love having this built-in detection, that sort of native integration of Dynatrace with these platforms. Cool. So to kind of close the talk, so from a product side,

I'm not sure how much you can allow us to look into the future, right? But we are at Perform, and we are always making announcements. Is there anything you can tell our customers what's maybe coming in the next quarters? Anything around obviously these technologies? Sure.

So we have already spoken about that yesterday in the hot day. So we've had like a hot day around OpenShift and Kubernetes where we gave like a quick roadmap update. So it's nothing new right now. So we will release a dedicated Kubernetes dashboard. We plan to release it in March, so it's really coming soon. And we will, of course, enhance further with these initiatives to have dedicated Kubernetes and Cloud Foundry dashboards.

We also plan to work more on Google topics going forward. So there is a lot to come. Service mesh is a hot topic. We see it everywhere. It's not only about Istio, it's also about Linkerd and stuff. We see customers using it for blue-green deployments and all that stuff. So this is also something we will address this year. Perfect, cool. Hey Alois, thank you so much. I think we'll go back and enjoy the show. It seems the background noise has slowed down a little bit, which means probably the people are running to the next sessions. Yes, it seems like that.

Okay, well with that, thanks Alois for pure performance and Andy Kretner.