PurePerformance - Automated intelligence for your multi-cloud IaaS platforms with Gary Carr

Episode Date: January 30, 2019

Gary Carr shares his experience with Dynatrace IaaS cloud support and a deep dive into how Dynatrace open AI provides intelligence into the capabilities and technologies of Azure, GCP and AWS.

Coming to you from Dynatrace Perform in Las Vegas, it's Pure Performance! Hi everybody and welcome back to Pure Performance and PerfBytes coming to you from Dynatrace Perform 2019 in Las Vegas. Once again, I'm Brian Wilson, and my co-host is the one and only Mark Tomlinson from PerfBytes. Hello, Mark. You know, there are other Mark Tomlinson's out there. There's one guy who was like... I said Mark Tomlinson from PerfBytes. Well, that's true.

That was a certain qualifier. But yes, I am here supporting this initiative, and I'm very pleased to be here in Las Vegas. Lovely Las Vegas. We're lucky to have you. For anybody out there listening who doesn't already know, Mark is the reason why Andy and I did this podcast in the first place. But enough about us. Let's talk about you. Enough about us.

Let's talk about me. Actually, Mark, we have another guest here with us now, Gary Carr from American Fidelity. Gary, how are you doing? Very well. How are you doing? Doing well. Mark, how are you doing? I'm great.

I'm glad we're going to dig in a little bit because I understand Gary has some experience using Davis, and I really want to hear more. But before we get to Davis, and I really want to hear more. But before we get to Davis, yes, before we get to Davis, Gary, you're from American Fidelity. Why don't you tell us a little bit about who you are and what you do? Okay, I am Gary Carr. I am American Fidelity's cloud infrastructure architect. And a little bit about American Fidelity is we're a private,

we held family-owned insurance company in Oklahoma. We have over 1,700 employees with over 2,500,000-plus policies in force with various products that we sell. Pretty decent-sized IT department. We have cloud deployments. We have on-prem deployments. And that's us. So, Gary, we see that the talk you just gave, you're using, well, to us, for Mark and I,

the most interesting part is you're using Davis. You're doing the integration with Davis. For people who don't know, Davis is where you can integrate Dynatrace into chat ops. This is typically through Alexa. Are you using Alexa for this? So I use Alexa. I also use, on my iPhone, I use the Alexa app. And I use online. So I've got the interaction online, Alexa and the Alexa app on my iPhone. Okay. And what do you, how did you, so you have

a bunch of monitoring, right? Like I don't want to, I don't want to brush over the other stuff, right? But you're, you're doing monitoring. You have a bunch of dashboards going on here, which I think a lot of people can relate to. What prompted you all to set up the Davis component? So when Davis was first being talked about, I had actually talked to Dynatrace, and I was going to become a part of that team to help do that several years ago. I had some issues in my family with family members that I had to help, so I could not do that. But I had purchased the Alexa and got that all set up to do that. I brought the GitHub down and was getting ready to do that. And then for that reason, I couldn't participate. So as that progressed and got better, I wanted to be a part of that. So as soon as we

got to a point where I could use it, I set up my Davis account. We integrated it in. I set my active gateway up so it could communicate. We worked on the firewall rules to get all that communication working. And then I started using it. First was Alexa, doing the Alexa demo, teaching people at the office how it worked, talking to it. And then as it progressed and as the features came out, then I wanted to try it on my cell phone because I wanted to be able to listen to my morning report, check problems on the way to and from the office, or just different times throughout the weekend. So set it up on my phone, got all that working. And so that's really where I started a long time ago

and wasn't able to really get involved in it. But now I'm involved in it and trying out new features. I'm running the beta version. So I've got the new failure rate and the new more contextual I'm running the beta version. So I've got the new failure rate and the new, more contextual aware engine running. So it's working very good. So you're usually mostly using it as chat ops, or are you also integrating it into, or sorry, that's voice ops, right? So I think I misused the terms before. That's more of a voice

ops, as they say. And then there's also like the idea of integrating it with chat ops, like slack are you are you doing the slack integration type of thing as well or is it mostly just voice at this point so at this point it's voice plus the interaction um online through the website where you can just simply talk uh type in text talk into it um i do have like my problem notifications go to team channels we We use Teams. We don't use Slack. So we're a big Microsoft shop. So they chose Teams. But we do have the interaction for Dynatrace going there.

I haven't brought full integration in with Davis into that yet. But you, of course, can from that add comments to your problems. And then those get attached to the problems. And of course, depending on whether you go from the voice side and you say add a comment to a problem, it goes, well, that also gets updated and pushed to the channel. So those updates will go. So I'm not fully integrated in all the chat interaction with Davis,

but we are continually building and adding new features to our environment. And we continually grow. And I'm not, I'm not asking this facetiously. I just know for me personally, I, I, I kind of find it weird talking to computer, but you've, you've obviously gotten over that. Was that ever a challenge to you or is it super helpful? Like what's, how has this made it? I know you mentioned something about on your commute in, you're getting updates. What has been the big benefit and how use have you gotten to talking to a computer, which to me just seems a little like too futuristic for me.

So I think most of us have grown up talking to our computers, whether they're listening or not is a different story. Now that they can listen, it's pretty interesting that they can actually take what we say, process it, and then bring back results and give it to us. So just the pure technology that it uses, to me, is really interesting. And, for example, one of the things in my broadcast is I was talking to Diana Davis and I said, play ACDC. And of course she does. Well, I don't know what you want me to do. Could you, you know, I can't do that. So it's interesting to try to play around and you know, the response that you're going to get, which of course in life is not like that. So you can actually give a question and get a response back, like shaking the eight ball

and you get the result back if you know what's coming back. Um, so I've enjoyed talking to that. I mean, I enjoy getting in the car and saying, uh, give me my morning report, which by default, it does, or give me a list of problems. Um, and it's kind of interesting to listen to that feedback. Now there is some, you know, there is some way that you have to specifically talk to it. It's in a learning mode. While it's really good, there are still some nuances about the way you talk to it and having to repeat a couple times. But overall, it's really good. And that engine, they've overhauled the engine.

It's a lot more contextual aware. So it's growing, and it's getting better. And that's what I like is getting exactly what I want to hear precise in a short amount of time. And then I can go back and listen to the radio or whatever else I want to do or call someone about a problem. Yeah, I think that's great because obviously this is not one of the main uses of Dynatrace, right? We did not design a product that was for people to talk to, right, on their lonely car rides into work. But it's a small little side piece that's in there, but suddenly became a central piece to your component. And I've always found, you know,

I work on the pre-sale side of things with Dynatrace, so I don't use it as much in a practical day-to-day usage. But I start seeing all these little nooks and crannies in the tool that have this potential to be used by people. And I shouldn't say surprising, but more often than not, I see people who discover them and then run with them because they're like, oh my gosh, you can do this. And this is just another example of it. Obviously, the bigger thing I think you talk about is doing multi-cloud, on-premise cloud monitoring and using synthetics and a lot of all these other pieces, which I want to spend a little bit of time on.

But I think it's great that you've embraced Davis, this kind of side feature that, I think when we first introduced it, people were like, oh yeah, I'm going to talk, some of our competition and some other people were like, oh yeah, look, yeah, you're going to talk to your computer, blah, blah. And like, now it's like, yeah, people are starting to do that. And were you at Perform last year at all or no? I was at Perform last year. Yes. So you saw the other new thing they introduced with the virtual dashboards, like the minority report type of thing. Are you

going to jump on that if it gets, becomes available? I mean, I want to try to do as much as possible. You know, some of it, some of it of course is, is how applicable is it to each company that's going to use it. So one of the things with, for example, with Davis, one of the original reasons we, that I look, want to look at it is because in our other building we had, we were talking about having a knock and I wanted the ability for people to talk and to be able to broadcast and do that. Well, we moved to a new building and now we have this big open structure. So having something that actually talks out in the open is distracting for a lot of people. So I was like, okay, well, I don't know that I can

use it the way I wanted to. But now that you have all these wearable devices and you've got the devices that interact with you wherever you're at, whether you're in the car, whether you're at home or whatever, and I can have that and get that information available to me, you know, we typically aren't an eight to five, go home and don't worry about the rest of the world. I'll look at it on Monday. There's real problems and real things that happen. And you have to use whatever tools you have to figure out what's going on. So having the ability to just at any point speak to my watch or speak to my Alexa at

home and get details like specific about problems or applications or user activity. I just want to see what's going on. And the new features that are going to be coming out and kind of working with the team and trying to move forward and say, hey, what else can I use it for? What information can I get out of it? And I haven't even got into the deployment side of it. into our full automation with the cloud and containers and really reacting to problems and scalability and being able to deploy stuff, whether it's automatic or whether I have to approve it and speak to it, that's kind of the neat stuff that I'm looking to.

So, you know, the VR and looking and touching and interacting with all these things, that new technology is going to be pretty cool. Again, it is, like you said, a lot of people in the beginning didn't want to talk to computers, but I find it kind of humorous that the number of devices that are purchased at Christmas and that are installed in even our parents' houses, and they barely know how to do anything really, but they can talk to it. They can ask what the weather is.

They can ask what the temperature is outside. And if they are capable of doing it, then it should be secondhand for us younger. And I'm 50, but I'll say us younger guys to do that. So you guys are doing a lot with dashboarding as well, right? I'm not sure if this is just part of the presentation, but it looks like you have a lot of pretty robust dashboards to set up a lot of different pieces of your monitoring. Again, I mentioned you're doing infrastructure, application service, database, synthetics, as well as your problems. So I can see the augmented, the AR dashboarding being useful at some point once that starts becoming somewhat realistic. But what are you finding with, you know, part of the appeal of Dynatrace is the ability to perhaps rely less on dashboards and more on the smart alerts from the AI. But obviously there's use for both of these components. Where do

you see the balance between just relying on the AI to tell you things, but also having a piece of glass with data upon there? Right. So, you know, at this point, everybody still wants to kind of get a, you know, we've got support teams and their goal is to support whatever it is, whether it's processes, whether it's servers or whatever, or users, they want to react ahead of time. So having a dashboard with having high level, you know, the honeycomb looking dashlets that show them the number of services. And you can easily see when things are red, when things are flashing, that there's a problem. You've got that. So we have the visibility through these dashboards that people

can run on TV screens now since we're, you know, we've expanded past our laptops and now we're pushing wireless feeds to TVs. The dashboards are really helpful for that. In addition to that, a lot of the people like different, not just technical people, but maybe management and executives, maybe they want to have a dashboard sitting somewhere so they can just keep an eye on it. They're not going to do anything necessarily with it. They just want to see what's going on. They want to see the numbers because when someone comes in their office and sits down with them and that's going, that's pretty cool. And they're not necessarily worried about the alerts and reacting to them, but they just want to show off this new technology that's out there and the dashboards are really cool and they continue to grow and new features are there. But then you go to the alert side and the problem side. And by setting up the different profiles,

we can filter and tag applications and hosts with whatever tagging mechanisms we want and filter those. And then that can go to a, for example, us, it's not Slack, but it's team. So we can have alert channels so the developers can get the alerts, not necessarily, they can get the emails, but those emails, you know, sometimes with all the emails and everything else you get, you want a different channel to be able to filter. So they can get that in their chat channels and they're talking about stuff all day in those channels and then they get alerts. And so now it's all in one thing. And as opposed to the email back and forth and responding, they're real time. They're sitting there, they can click on it, they can go straight to Dynatrace and get down into it, get down to the details, click over back to that chat channel,

communicate about it, have somebody ask a question, and it's all right there. And it's also documented in a time slot. So here's an alert. Here's the problem number. You can click and interact, but then you can have a conversation about it. And if you have to notify someone about a service or a database, you can get them involved. So it's kind of nice to have both to me for different reasons and different purposes. Yeah, and I think you're totally right there. I think some of the initial idea is way back where we can start moving on, but there's obviously a lot of use cases for them. I did have one question that you were talking about. One of the great reasons is to have that proactivity,

where you can start seeing things going wrong. You can jump in before it actually becomes a problem. Have you all at American Fidelity started looking into using the api integrations to move a little bit towards no ops or at least for common problems that you're able to that you know that might crop up taking your playbooks and automating them based on data you're extracting from dynatrace so that you can remove some of that um being proactive you can automate some of that proactivity. Has that come into your scheduled ideas yet? Or I don't even know how to say it, but you know what I mean.

Is it part of the plan, bigger picture? Definitely. I mean, we have talked about it, and of course we want that. You know, one of the big deals of moving things to the cloud is that work workflow and automation is you want to be able to now that i have the ability to set scalability and thresholds and based on this i want it to automatically perform this action so we can do that or you can do it even through dynatrace and say here's some thresholds and when this happens i want this to go run this script and i want this to push this out and i want it to expand it. And when it gets to a certain

point, I want to bring it back. So definitely we want to do that more. Right now we do a lot of, we are big in DevOps. So continuous integration, continuous build, continuous deployment, integration with ServiceNow. So we've got controls around that. But as far as a no ops, we are not a no ops shop yet. So it will be nice to get to a point where that is fully automated, where a build can occur. Dynatrace can watch it. It can check for things and it says, okay, it's good. Now we can move on to the next perspective.

It's here. All these checks and balances are in place. Let's go ahead and deploy this out. So yeah, that whole op side of Dynatrace is a side that I've really been wanting to get into in the last probably two years. But there's so much else going on. We are really growing a lot in our cloud,

mainly in Azure. We've got some AWS stuff we're doing as well, trying to decide where we want things to run and what cloud and why. Why do we want things to run on-prem? Like the hybrid connections, what do we want to run on-prem, but we want to make use of GPU stuff out in the cloud so we can burst something up and go run some stuff on it, be done, come back and still be processing stuff, making use of storage so we're not paying additional capital costs on additional storage. Those are all the things that we're really looking at. And then when you look at all that, you have to look at automation. You have to look at making sure things are spun up and spun down so you can be efficient with your funds.

Because pushing to the, moving to the cloud can be way more expensive or it can be way more efficient. I mean, if you do it right, then your cost overall should be pretty smooth. But when you go calculate a bunch of storage in the cloud and say 24-7 for all year, you're going to be like, oh my gosh, we can't do this. Well, when you say, okay, how much storage am I actually going to use? What's the growth? And then is that storage compute and everything else? Does it have to be available 24-7? Well, most of the time it does not unless you're a 24-7 upshot. But if you are, you've got revenue coming in and it's going to pay for that.

So I think you have to look at the whole picture, which automation is definitely part of that. Yeah, and I was going to say those cloud costs can really scale up quickly if you give developers the autonomy to just deploy as they need. Right? Because they're not seeing

any part of the finances. It's like, oh, I need another server. Let me throw 50 gigs on that one and get five CPUs. It'll be really fast. Five CPUs.

Who does five? I don't know where I got that number from. So in terms of the no-ops, hopefully, I know there are several people here uh at perform now uh doing some no op stuff hopefully you'll get a chance to chat with them

uh maybe they can uh share some of their experiences with you uh if not we can see if we can hook you up with some of them um because yeah there's definitely a lot of people or some people are definitely exploring that now so be good to catch you connected with them. Before I let you go, I need to ask, so is this your first time in Las Vegas or no? No, in fact, I was there for reInvent not too long ago. Okay. And it's the new year, right? So I got to ask,

do you have any performance-related New Year's resolutions for 2019? For me, mine is more synthetic licenses, more host units, so we can continue to grow. We can continue to expand. I really want to get more of our full automation in VM deployments. We are going to be moving more into containers over the next year and a half. So I want to get that where when those containers are pushed out, we are automating all that process has all the different components look at different things, and I don't have to go into AWS and have that separation.

I like the way Dynatrace gives me the view for both of them, and then I have that full stack so I can see everything that's going on. So I really want to really work on what I can do to help get us further down the road so that we have that full visibility. Because every time the portal changes, you know, now you got to re-figure out where you were clicking. Okay, this was on this page. This was on the third page when I clicked through. Where's those features at now? So I think that's really where I want to get is, in addition to that, is more training and deployments for our teams and getting into the no-ops model so that we can automate the growth. So just the smallest. Just the smallest.

And that I hope to do by June. So I wanted to thank you for taking the time to talk to us today. We hope you're enjoying your time here. And good luck with everything in the new year. And we'll see you. I'm sure we'll see you around the floor here somewhere. I'll see you around. Thank you.

Thank you. Thank you.