PurePerformance - Automate your monitoring strategy with AI powered Infrastructure Monitoring with Adam Dawson

Episode Date: January 30, 2019

Dynatrace's Adam Dawson gives a few tips on gaining visibility across your entire environment, including infrastructure-only nodes with a single, all-in-one solution. His session showed how to extend Dynatrace to monitor your cloud infrastructure health, in addition to your application performance.

Coming to you from Dynatrace Perform in Las Vegas, it's Pure Performance! Hello everybody and welcome to Pure Performance and PerfBytes coming to you from Dynatrace Perform 2019 in Las Vegas. I'm Brian Wilson, and I don't have my co-host right now, Mark Tomlinson, because, well, he's occupied right now. Anyway, but I do have someone very interesting, Adam Dawson. Adam, how are you doing today? Do you want to introduce yourself and let people know who you are and why I might be talking to you?

Sure. Hi, Brian. How are you? It's good to be with you. I'm Adam Dawson. I'm a director of product management here at Dynatrace. And I'm here to talk with you about one of Dynatrace's best-kept secrets, cloud infrastructure monitoring. Yeah, so you just had your presentation, right? You said your breakout session, and I'm looking over your slide deck. And I kind of laughed, not in a bad way,

but laughed at the, I thought it was a little novel, the idea is the best kept secret of Dynatrace infrastructure cloud monitoring. Because if you think about the big picture of Dynatrace, infrastructure cloud monitoring is not one of the first thought of or primary things people think of. And I think what we're finding a lot during Perform so far

is there are a lot of ways outside of what people think of typical APM that people might use Dynatrace. A lot of people think APM tools is like deep dive monitoring your container and the code inside and everything. But we're discovering all the different things people are doing with their APIs. We're discovering people using chat apps, voice ops. And now you're here taking it, bringing it kind of almost back to the basics of just the infrastructure and cloud monitoring. So tell me a little bit about what that is in context of Dynatrace or in monitoring, first of all, but also in terms of Dynatrace and what you were covering in your topic. You know, it's a very interesting topic because, as you mentioned,

it may be something that seems like something that you would have talked about years ago. And you've heard a lot here at the conference about deep dive code level visibility and chat ops and voice ops and IoT and the next generation of monitoring. But what we've really found in the last few years is that as applications and infrastructure become sort of inextricably linked or inseparable, the advent of infrastructure as code and software-defined hosts and networking and storage, that some of the traditional tools that customers and enterprises have been using to monitor their infrastructure environments really aren't able to keep up with the shift to a hybrid cloud and to a mix of public

and private cloud and data centers all the way to microservices and serverless, right? And so what we found though is that, and the reason that I describe this as maybe the best kept secret at Dynatrace is that in some cases, I think customers have found it on their own and have realized as they move to the cloud that the value that Dynatrace provides, not just for the application, but also for the infrastructure layers and the hosts, all the way down to the process level and the CPU and the memory and understanding what's going on in their environment and how it impacts customers is really very strong. And they're able to use Dynatrace as an all-in-one solution to monitor everything from the application all the way down to the host. Right. And so to clarify, in my mind at least,

what we're talking about is basic infrastructure monitoring. We're not basic in a sense because now it's gotten more complicated and more difficult to route things and get everything all in one. But you're talking about the infrastructure monitoring. And I think a lot of people now are so focused on, let's say, containers or their runtime and all these pieces that there's always that infrastructure group is looking at the infrastructure and all. We don't really care about that. But it's not like anything has changed. It's just the location of the servers has changed.

You still have to worry about a full disk taking down your application. Or maybe if someone's writing extensively to logs in their application, that's chewing up the CPU on that server, right? And therefore, that's going to have an application impact. So the infrastructure monitoring pieces is still obviously just as important. It never goes away unless you go away from infrastructure, if you're using serverless functions and something like that. But I guarantee you, if you're using Lambda functions, AWS is monitoring that stuff, right? Someone's monitoring it somewhere. So the interesting thing here is with Dynatrace,

if you are running the one agent on your application, you're automatically getting that infrastructure monitoring. The one agent goes from the infrastructure all the way up to the application, all the way out to the end user. But there is another thing, and I think this is one of the things that people might not know so much about is, let's say you have an application running that you can't put an agent inside, a deep monitoring agent inside. Maybe it's a closed application or maybe it's some antiquated technology that we can't get

inside of. Is there still a way for us to monitor that piece of infrastructure? Yes, absolutely. I think it's a perfect example of closed application, right? Or even what we might call a business application or an internal system where you don't necessarily need or cannot do that deep code level visibility. But with the Dynatrace One agent in a mode called infrastructure-only mode, which is just a toggle switch that you can flip inside the Dynatrace platform, you can still get the visibility to all of the CPU, the memory, the network connectivity, and also the process level diagnostics for how much of that CPU each process is consuming, process-to-process communication.

And really, the power of it is not just the infrastructure monitoring itself, but also Dynatrace's SmartScape technology, which builds that device and that infrastructure host into the end-to-end topological view or the map so that it can be used by Dynatrace's AI engine to perform root cause analysis and remediation and help you understand the real impact of a particular part of your infrastructure on your application and on your end user and help you understand when that's the problem, right? So it's not just the ability to monitor infrastructure, but also the ability to apply the Dynatrace

AI magic, as it were, to the infrastructure part of your environment as well. Yeah, and I think that's a really important point, right? That the AI part is, AI is consuming all that. It's not just looking at code level transactions running through. It's putting all the pieces together from infrastructure up. And by including that infrastructure-only agent for those cases that you can't do the full agent, you still get the benefit of the AI on those components.

Now, just in terms of, you know, not that we're here to talk about licensing or anything, but in terms of, you know, I imagine if there was any Dynatrace users listening, they might think, well, you know, if I have a license for an agent, is an infrastructure one consuming an entire agent? I believe it's a completely different pricing

model for that, right? Yeah, absolutely. But the good news is, it's really easy, right? When you have a Dynastrace license, you typically buy a pool of host agent licenses, you know, maybe 50, 100, 1,000, whatever. And when you deploy those host agents on servers, they consume, you know, one or more licenses depending on the size of the host. If you choose to use infrastructure-only monitoring instead of full-stack monitoring, on that host, you only consume 0.3 or 30% of the same number of host units that you would consume if it was a full-stack agent. And the good news is, like I said, you don't have to buy a different SKU or a different pool of host agent licenses. It's all from the existing pool that you have, and you can toggle them on and off as you wish in the Dynatrace platform to meet your needs.

Okay. And I should also point out, it's the same agent you deploy, you're just going to set a flag to say, be infrastructure only. So in terms of, oh, do I have to get another thing and know what to deploy where? No, the great thing is you're still using that one agent concept where you deploy the same exact thing. And all you do, especially when you're automating your deployments, is you add that infrastructure only flag to true. Now, there's another cool thing I think you were talking about in your presentation, which was sending out some of this information to external sources like ServiceNow, like your ITSM and CMDB components, right? Can you tell us a little bit about that? You know, you've probably heard of this in different sessions and through the years at Dynatrace, but we really have a strong set of integrations with common and popular

sort of IT service management systems. And when Dynatrace is monitoring your infrastructure, as well as your full stack hosts and your applications, it's very easy to go into Dynatrace and set up an integration with a system like Jira or with email or HipChat or Slack to send those notifications or those alerts to any sort of system that you like and several more, not just those, several more on the alerting side. But also our integration with systems like ServiceNow CMDB really gives you the opportunity to keep that CMDB up to date and manage the CMDB in real time rather than the typical way of managing a CMDB, which always seems to be out of date. So what we call this is sort of CMDB integration and enrichment. So you can really enrich that CMDB with information from your SmartScape topology and keep it up to date in real time.

Right. So what we're saying there, I think, and correct me if I'm wrong, because there's always so many components to Dynatrace. It's always fun to wrap your head around all the different things that you can do with it, right? But with CMDBs, typically people have to enter in their known systems, right? And I believe it's mostly like either you import it from a spreadsheet or something, but you have to know your system and enter all that stuff into the CMDB, I think. You can correct me if I'm wrong there. But what we're saying is with Dynatrace, with all the auto-discovery, and if you're not familiar with the smartscape, it's the auto-discovered interdependencies on host infrastructure process, application, all those levels. We can feed all that to the CMDB so you

don't even have to know what the connections are. In most cases, people I don't think do know what all the connections are and everything that's out there. So that's a real, real great benefit for that, which then ties again, you know, the CMDB brings us all back to your infrastructure monitoring and all that. Sure. Yeah, absolutely. And the other thing about it is it's not just that keeping your CMDB up to date is difficult, but again, as the world has migrated to a cloud infrastructure,

which tends to be dynamic and fluid and comes and goes, I mean, really managing CMDB manually is going to be impossible, right? And so this sort of integration to inform your CMDB with updates in real time is really powerful for users as well as enabling, right,

for keeping those existing sort of tools and processes and best practices that your organization has developed in place in the cloud world. Excellent, excellent. Anything else you wanted to bring up about your talk that you just gave? Well, you know, I thought there was just one more quick point here that we always want to cover when we talk about this is I think we spent a couple minutes talking about some of the legacy tools that people have used over the years to sort of do infrastructure monitoring. And I think as you've seen a shift to cloud, you see almost a very similar approach out in the market that you saw in the legacy environment, which is a point tool or a specific tool for each technology that's designed to monitor the specific kind of metrics

or parameters of that technology or that deployment. And I think at Dynatrace, our point of view is that we don't really want people to repeat the mistakes of the past and end up in this environment where there's far too many monitoring tools that each have a very siloed view of what's going on and don't connect the view of what's going on with their part of the environment to the overall application or user or business impact. And also don't have any context about what's going on to make the monitoring and alerting and integrations intelligent about what you're doing. And so what we really see, I think, as part of the power of Dynatrace, whether comparing to these legacy tools that you might be ready to get rid of, or some of the newer tools that you see out that are designed supposedly for the cloud, is that so many of them still just want to present you, I think, with a lot of data and a lot of charts and a lot of information that lacks context about what's going on in your environment.

And so what we really see, again, is the power here for extending Dynatrace to your infrastructure monitoring, is not the unique ability to see every specific metric of every specific technology, but really the unique ability to understand how a certain part of your infrastructure impacts the rest of your environment and your end users and your business. And so I think that's why customers have found that the monitoring that Dynatrace provides for infrastructure really has been very powerful for helping them manage their environments more effectively and prioritize and scale because Dynatrace helps you sift through all that noise to get really actionable distilled data and also keeps those baselines and thresholds up to date

in real time automatically without you having to go configure 10 or 15 or 20 different tools and have 10 or 15 or 20 different teams with a different view of the truth. That's an excellent point. I think it goes back to the idea of the difference between data and information. Data is just numbers. It's just data.

And information is when you take that and turn it into something actionable, when you give it context, when you put it all together. And what you're saying there I think is a really good point too, about if you have a bunch of these different tools together, you're getting a bunch of data from them. And then you have to correlate the data between the different tools to try to piece everything together. Whereas when it's all being fed into the Dynatrace AI engine from all the different layers, we're no longer relying on correlation. We have direct causation because everything is known to being in that. And I think that's one of the tricks with modern tooling is having a causation approach instead of a correlation approach.

That goes back to my old, really old days back in testing where I used to be a load tester and performance tester. And we'd run a test and we'd say, oh, there's some errors, there's a problem. And the developers was like,

oh, what time was that at? And then they start going to look in the logs around that time to start looking for errors again. But again, that's a correlation because just because that error occurred doesn't mean it was from that and looking at the CPU, but it gets all complicated and ugly but that was the old way

of doing things but we're seeing that extending into so many of these places where there's multiple tool sets it's just you know same problem but just a different skin on it um and i think this is a great approach yeah and everybody's been in a war room right everybody's been in a war room where the database person is pointing at the developer and the developer's pointing at the operations person and so on and so forth, right? And they all have perhaps a log or a tool or a view into what's going on that tells them that it's not their fault or they don't know whose fault it is. And I think that the value of bringing all this into one interface that can be shared and traced end-to-end and automatically configured and baselined and thresholded so you know exactly what you're supposed to care about versus not care about.

And everybody has the same view of the truth is actually a very powerful component as well. Maybe you would view that almost as the value on the people side as well as the technology side. Yeah, it's funny. When you were saying that, I was thinking it's the, I don't know who peed in the pool, but it wasn't me problem. Yeah, sure. I don't know. It just struck me. All right.

So is this your first time in Vegas, I got to ask? I've been to Vegas a few times. I actually was at Perform last year as a guest. I joined Diatrate shortly after, so I'm excited to actually was at Perform last year as a guest. I joined Dietary shortly after, so I'm excited to be back at Perform this year. It's a great venue and a great conference.

Well, hopefully you'll go home with the same or more money than you came with. No. No, you're like, no. Absolutely not. And one question we're asking everybody is,

since we're at the new year, do you have any performance-related resolutions that you would like to state? Oh, you know, I'm not much of a resolutions person, so sorry, I don't really have anything for you there. That's okay. I think we had one other noncommittal answer, and we'll just put you in the performance is unimportant pile. I'm kidding. You know, all the gym rats always complain about all the people who show up in January after the New Year. Oh, yeah. So I'm a guy. I like to wait until February or March to get started so they don't think that it's just a New Year's resolution.

So call me back in a month or two and see how I'm coming on my performance-related resolution. Awesome. All right, Adam, thank you very much for taking the time to talk to us today, and I hope you enjoy the rest of Perform. Ryan, thank you for your time. It was good being with you. Thanks. Thank you. Take care.