PurePerformance - Perform 2020 Andi on the Street: AIOps, Performance Engineering and Deployment Strategies

Episode Date: February 5, 2020

Andi Grabner, our man-on-the-street, gets the scoop on:-Leverage AIOps with Dynatrace with Wolfgang Beer-Load & performance engineering as a self-service ​with Rob Jahn-The right way to deploy canary, blue/green and feature flags with Safia Habib

Coming to you from Dynatrace Perform in Las Vegas, it's Pure Performance! Hi everybody and welcome to Pure Performance and PerfBytes, coming to you from Perform 2020 in Las Vegas. Our man on the street, Andy Gravner, has sent us a few interviews, which we've compiled in this episode. There's a short musical interlude between each, so be sure to stay tuned in. Take it away, Andy. Welcome, everyone, to another episode of Pure Performance Cafe. Today, again, live from Las Vegas, from Perform 2020. And I bumped into another colleague of mine, Wolfgang Beer, product manager at Dynatrace from the Linz Lab, but today we are, well, in Vegas as I just said.

Hey, Wolfgang. Hey, Andy. How are you? Good, good. It's a little stressful here. As you can see, running around, sweating a little bit, but that's what it is, right? There's long distances here in the Cosmo. But Wolfgang, I wanted to catch you before you actually

go into your breakout. I think that's coming up soon. You have a breakout around AIOps explaining how people can leverage Davis, our AI engine. Can you elaborate a little more on what people are going to hear, especially those that cannot make it live, maybe watch the recording? So who is there with you? What are the key takeaways? Yeah sure, so we will jump into our session pretty soon. So we, that means Jay Cotton from Kroger and my person, and we will talk about how you can leverage the power of Davis and AIOps built into Dynatrace and how you can automate or reach a new level of automation within your AIOps departments in order to automate

operations within your company. So as Steiner Trace and especially Davis has a lot of context about the ongoing operations and the services that are deployed in your environment Davis can help you a lot in order to automate various tasks within your operations departments. Very cool. I mean, I know we've been doing podcasts together, let's say performance clinics together, where

you explained the approach of our, let's say, deterministic AI. And I think we're also very particular now that we just don't call it an AI, it's a deterministic AI because we do have the dependency tree that we call SmartScape. So we can do fault, what is it called, fault tree analysis? Fault tree analysis, exactly, yeah. And I think, I mean, for those people that may just hear the buzzwords around AI, AIOps, can you quickly explain what you see as the product manager as the key differentiator of our approach to the approaches of other solutions that are out there that also claim to do AIOps? Yeah, it's really about automation. So Dynatrace, especially Davis, was built in right from the beginning

in order to support fully automated processes. So Davis has the capability of performing a full stack fault tree analysis in terms when a problem happens in your environment and Jay will Jay cotton will then explain in our session how that helps or helped Kroger to reach a different level of automation in their environment yeah and I think for me I mean I've been we both have been working for Dynatrace for a long, long time. We both came from the Appmon times, right?

So, also I know from the Appmon times, when we basically, when people were doing root cause analysis, you started at some place and you drilled down, you basically walked a certain path, and what you really did, you walked along the pure path dependency tree, basically. That's what we had with Appmon. Now with Dynatrace, we do not only have the dependency tree

along the transactions, we we also have as you said full stack meaning yes I always call it the vertical and horizontal dependency map right and we walk through the whole tree. Exactly so so basically right from the beginning Davis was was actually called Andy in a box, if you remember that. Yeah, I remember that. And meaning the purpose was that within modern service infrastructures, microservice infrastructures, it's nearly impossible to follow the pure paths and the transactions, the verticals and the horizontal stack manually anymore. So infrastructure changes too quickly for humans to catch up.

And that's also what Jay is explaining in this part of the session, that you can't scale human ops teams as you can scale modern infrastructures. And so there is a tremendous need of automation like Davis in order to catch up with the technology today. Now, Wolfgang, I also know you've been a main driver behind all the APIs. And obviously I assume the APIs of Davis play a big role also in the way Kroger is using it and others are using it. Are you also touching on the APIs a little bit or is there in your session?

We are basically touching the capability of how you can ingest and leverage the external event possibilities so to enhance the information that is coming in from Davis and how you can ingest your own information into the Davis fault tree analysis. Perfect. Maybe last word, I'm not sure if you know, well you should probably know, but Kroger coming back to Jay, you mentioned that they have reached a new level of automation. Any numbers that you can point out because people are always kind of excited when they hear numbers like, you know, that many, I don't know, incidents removed, that more uptime, anything that you can tell us about? I have no concrete numbers. You have to watch Jay's session afterwards.

But he told me that they have a neat ops department with around four people working in it. But I asked Jay about the details about that. And he said that it was simply impossible to scale at that level where their infrastructure and service infrastructure scaled. So that was the main driving reason behind using Davis to automate. Yeah, perfect. And actually, thanks for the answer. It's a perfect plug that we don't give away all the secrets of the session, but actually watch it.

Awesome. Well, Wolfgang, I want to release you now so that they can catch your breakout. And I'm pretty sure I'll bump into you at some point later this week or maybe at a bar at some time and actually celebrate the success of the conference. See you later. See you. Bye-bye. See you.

Welcome, everyone, to another episode of Pure Performance Cafe, live here in Las Vegas at Perform 2020. And I bumped into another colleague of mine. It's awesome. I mean, there's a lot of colleagues here, but especially one that is also presenting in the track that I'm track captain of, Release Better, Softer, Faster. And it's Rob.

Hey, Rob, how are you doing? Hey, Andy. Great to be here. Yeah, great that you... Well, thanks, first of all, to actually kind of take the challenge of doing a presentation. I know it's a lot of people here, but I think there to actually kind of take the challenge of doing a presentation. I know it's a lot of people here, but I think there's a lot of great content we need to deliver to our customers.

Now, Rob, before we get into the details, maybe for folks that don't know you yet, can you quickly explain your role at Dynatrace? Absolutely, yes. I joined Dynatrace a year ago. I'm called a technical partner manager, so I work really closely with our strategic partners such as AWS and Atlassian. And I go to a lot of different customer sites and meetups and really kind of talk about all the things we're doing with Dynatrace and a lot of the use cases and DevOps and performance engineering. And it's just been an exciting time to be part of the movement to really go to no ops and a lot of automation. And you have a lot of background in performance engineering in your previous jobs. I know we actually got to know each other years ago when you were still working

for another company. And we actually did a performance clinic on integrating performance into continuous delivery. And so that's also why, you know, I think it's great that we finally have you on board at Dynatrace. We used to work together in the same team for a while. And now, as you said, you know, you're focusing more on the partner side. But Rob, let me ask you and jump into the topic. So your session is called

Increase Quality and Agility with Load and Performance Engineering as a Self-Service. So the whole continuous performance, performance engineering as a self-service is a big topic that we've been pushing for a while. For folks that are still contemplating which session to attend or contemplating, Angella, watch the recording of your breakout. Can you quickly give a quick overview of what are you going to show? Who is going to present also from a customer side? What are people going to learn in your session? That's right. So, yeah, so you were correct. I have a long history of performance engineering.

So for me, this topic is really something I can relate to and I believe in because, you know, a lot of companies that are moving from traditional big applications, waterfall, and as we've evolved into offshore teams and centralized teams that do the performance test execution, that is turning out to be really a bottleneck and not getting enough performance coverage. And it's putting risk into our system. So the topic is really about how can we shift left all of those performance activities? How can we automate that analysis so that we can put these things into

software delivery pipelines that are automated? So it's really kind of excited that, you know, that's kind of where industry is going. That's where we need to do that change so that we get those feedbacks, push this back into the developer hands, but that expertise that all the performance engineers like myself and a lot of great people have the expertise to analyze, establish service levels. And so we want to leverage that expertise and really create, you know, a framework that we can enable the development teams to do that. So part of what we're going to be talking about is sort of that need and what's driving

our industry, you know, as we move to the cloud, as we start to have more and more services, more and more automation. That's kind of really why we need to do these things. And really, we're going to get into how Dynatrace, the platform, has a lot of out-of-the-box features to help make that possible. We have a very rich API that gets into all the time series data. We have the AI engine that can be accessed to automatically detect problems. And then we brought in Neotis, which is one of our testing partners. And so they've actually built integration into the product that kind of takes advantage of these features. So we're going to have Henrik, who's one of their evangelists, global evangelists all over the world that talks about their product. And then we're going to be followed by Panera Bread, who's one of our customers who has been going through this journey to transform and build this framework. So they've established a foundational observability platform that brings

in data from Dynatrace, from production that really establishes the service level objectives and targets. And they've really transformed their team to really do what I was describing just a second ago, where PE is now the team that does the heavy lifting to establish the workloads and then delivers a platform and the guardrails for various engineering teams to do self-service. but know that they're doing it with established targets and workloads that represent production so that the quality of those tests is there, but then they get that feedback faster, which is really why we titled the track Performance as a Self-Service. That's pretty awesome. I mean, there's a lot of stuff that you're going to talk about. A lot of stuff. Yeah, but it's exciting. It's exciting stuff because that's where PE has become more and more important. We know that users of websites, business customers expect high performance, high availability.

So it's not just a trend. It's just the reality that sites need to be available, performant, because end users have just very high expectations. And at the same time, we're going into a world that's more and more complex. You know, we have distributed applications that are not just on-prem, but now we have hybrid cloud. We have this big movement to containerized microservices. We have lots of teams now developing, you know, in parallel.

It's not just one big monolithic app. It's distributed teams. So the complexity is there, and that's really where, you know, we want to showcase, you know, you know, the real, the real power of Dynatrace to, to be able to work in a dynamic environment, to be able to automatically detect these changes, see the dependencies between things and really leverage a lot of the, you know, the long history of Dynatrace, which is really, you know, what inspired me to join Dynatrace. As a performance engineer, I used the Dynatrace platform in my work for many, many years as a performance architect with a consulting company. So I know, you know, firsthand that the tool is invaluable for, you know, performance engineers, as well as development teams and operations to see it.

And that's really, you know, hopefully this is, you know, one piece of the puzzle in a big organization, but, you know, where we can all use the same tool, have the same source of truth, where we can see everything from the front end to the middleware to the database SQL calls all in one tool is just a fantastic tool, you know, to complement something like a NeoLoad, where it will drive the traffic, you know, integrate with our tool. So now we have, you know, the best testing tool, We have the best observability tool, as well as the intelligence to detect the problems in an automated way, leverage these APIs to really make this automated process that can scale.

Pretty awesome, Rob. I think that was a great testimonial, obviously coming from a Dynatrace fan and an employee, but I'm pretty sure we'll also hear it from Prashant from Baneva Bread, as you said, right? He's always also been a long-term Dynatrace customer helping us also to improve our product, to evolve and to make the life of performance engineers easier.

That's right. And so he will have a lot of great insights too. And that's kind of where I would definitely come, not just for the technical parts, but really his insights of what it takes to transform a team. Because it's been a journey. He's had, you know, some fantastic success. He has a cross-functional team. And that's one of his, you know, secrets to success.

It's not just a single role. together, you know, with expertise from the ability to analyze the workloads to the deeper performance to obviously the software engineering groups, which are the bigger set that we're trying to support. So all working together kind of with the same source of truth, the same process and the same platform. And that's really what Prasanth from Panera is going to share with us, as well as what he's got planned ahead. So it's one of these continuous improvements.

So he's got much more to share about what he's doing next. And hopefully that'll inspire everyone as well. And they can learn from that. Awesome. All right, Rob. I know we got to get going here because the show is continuing. So all the best for your show, for your breakout. Folks that are listening in, make sure you either catch it live,

you still have time to walk over to the room or check out the recording. And Rob, do me a favor, if you have a chance, either you or Prashant or Hendrik, just take all of them and walk them over to the Pew Performance and the Proofbytes podcasting station, right?

I think it's great if they can share directly with Mark and with James and also with Brian that are sitting over there. Sounds great. Well, thanks, Andy. We'll see you later in the show. Yeah, see ya. Bye. Hello and welcome again to another episode of Pure Performance Café, still live in Vegas,

believe it or not. It's been a long week already, but we're not dead done, so there's more sessions to come. And great enough, I found another colleague of mine who is about to go into her breakout. And I'm actually very happy that Sophia is actually presenting in my track, the Release Better Software Faster track. But before we dive into the session, hello. Hi, Sophia. Great to finally see you. How are you doing? Hey, Andy. Great to be here. It's been exciting to be here and then meet everyone. So looking forward to doing the presentations. Yeah. Hey, how long have you been with the company? I have been with the company

for almost five years now. Five years. And I remember you used to live in the US and now I believe you moved to Canada, correct? Yes, it was indeed like that. So I used to be in the Waltham office. I went there every day and then I moved to Canada, I think almost a year back. Yeah. Well, and well now obviously escaping the cold and now you're in Vegas. It's definitely a little warmer here, right? I know. That's such a nice time that we have Perform Conference. It gives us the very much required break from the cold of Canada. Yeah. Hey, diving into the session that you're doing, I think it's called Reduce Risk and Iterate Fast,

so the right way to deploy canary, blue, green, and feature flags. Now, this is a long title, but I believe it's very important that it actually includes all of these key terms that are focused around, I think what the industry is also calling progressive delivery. Now, Sophia, what I would like to learn from you, especially for those people that cannot either attend your session live or are not here at all,

but they want to potentially decide on whether they want to look at your recording, can you give us a quick overview of what are you going to cover in that session? Yep, certainly I can do that. So as you said, it is a long title, but then there are a lot of key terms that need to be mentioned in the topic itself so that people can make a better choice of whether they want to listen to it or not. What we are trying to do here is when everybody talks about monolith to microservices, there's a lot of buzz in the industry. But just like that has happened, there is also the change in deployment models. With monolith, it used to be the big bang delivery or everything needs to go at one time. From there, we moved to a progressive deployment model, and that is what we want to cover in this session.

Talk about how progressive deployment models are available, what models are available, including Kennedy, blue-green, feature flags, and then talk about what's the right way to do them. And along with that, we also have one of our customers who's doing it currently and using Dynatrace to get value out of it. We'll talk about how it can actually be achieved in today's world. That's awesome. What are they using? Are they using Blue Greens or what are they doing? Yes, they're not using blue-green. They're using Kennedy deployments because that is the most risk-averse way of doing things. And they're a financial institution. So it makes a lot of sense for them to reduce the risk, not just iterate faster, but also reducing risk is a bigger part of their software solutions.

That's awesome. And obviously risk is very important, especially if you feel safer. I think you also, as an organization, get faster into the mode of actually being, let's say, more experimental friendly or more, how would you say, it's easier, right? If you know that you cannot necessarily break things in the bad way

because the system itself takes care of reducing the risk, then you may start actually trying out new things that you would have not tried otherwise. And I think that's very important. Exactly. That is what they are doing. Now they have become so comfortable in actually using those deployment models that they have faster deployments that happen and then they are able to iterate faster. But the first step was reducing the risk that they had with their architecture. That's perfect. Hey, and so if folks that are attending your session,

you said you give them an overview of all these different deployment models, and then I would obviously assume you also show them how to correctly set up or configure Dynatrace so for dynatrace also to then you know detect these types of deployments and so can you tell me a couple of things that people will learn yes so there are a couple of things that people will learn first they will learn an overview of all the different deployment models including what is blue green what does it offer what are canadianments? How can you actually do them effectively? A-B testing. And also there's a lot of talk in the industry about feature flagging.

There are a bunch of companies that allow you to do feature flagging. So we'll be talking about how there are software products that can enable this journey for you. And then finally, we'll talk about how you can integrate all of this with Dynatrace so that whenever you do a progressive deployment model, you need to be able to measure it. There has to be some kind of a quantitative measure that goes into deciding whether it is good enough or not. So we will talk about how you can actually use Dynatrace, be it tagging within Dynatrace or pushing events into Dynatrace using the APIs that we have, the rich APIs that we always talk about being the API-first company. So we'll talk about how you can use all these different features within Dynatrace

to ensure that you can actually quantify your blue-green or your Canadian deployments. That's cool. That's very good. Yeah, I mean, as you said, it's APIs. We're big on APIs, and it's even better than if we tell people how to use these APIs for the different use cases. Awesome. Hey, Sophia, I don't want to keep you any longer. I know it's about time for you to get ready.

For folks that are listening in, make sure you check out the recording of her session that is called Reduce Risk and Iterate Faster, the Right Way to Deploy Canary, Blue, Green, and Feature Flags. Sophia, thank you so much. Thank you so much, Andy, and I wish you a good rest of the conference. Thank you, you too. Bye-bye.