PurePerformance - Dynatrace PERFORM 2019 Power Dashboarding and More with Ben Rushlo

Episode Date: January 30, 2019

Conversation with Ben Rushlo, VP of Services at Dynatrace, talks to us about dashboarding in Dynatrace

Coming to you from Dynatrace Perform in Las Vegas, it's Fear Performance! Hey everybody! Welcome back to Dynatrace Perform! You can actually hear Brian Wilson on his own podcast now. Yeah. Because everything is working. Welcome. Back to another live interlude. Interlude?

It's an interlude. I guess. This is more of an intense. Well, there's two ludes and it goes between them. That's inter. Okay. If you were in the same lude, it would be an intralude. I kind of follow you.

And if it was every other lude, would it be a bilude? Biludial, I think, would be an intralude. I kind of follow you. And if it was every other lude, would it be a bilude? Biludial, I think, would be the proper term. So speaking of biludial, we have Ben Rushlow, VP of Services at Dynatrace. Really famous interludial person. I don't know what I was getting into here, guys. Ben is the Vice President of Services here at Dynatrace. Technology is easier with humor. Yeah, it is.

Yeah, exactly. So you can share some interesting stuff about your session And also the hot day, you did a hot day, right? We did a hot day, and that was pretty cool So the hot day was really well attended It was on web performance optimization Which is a very vague title But kind of split it

It's also a category There's a whole WPO web perf Steve Souders et al Hey, velocity day Shout out to velocity They al. Exactly. Hey, Steve Souders, if you're listening, I'm blocking you. Shout out to Velocity. They don't have that anymore, do they? But it was interesting because we weren't quite clear on what the attendees were hoping for. So we sort of split it down the middle where I did a half of it kind of on best practices around real user monitoring and how to get the most out of that,

which a lot of our customers aren't kind of taking advantage of all the tips and tricks and pro tips. And then Doug Pack and Joe Hoffman, who are sales engineers, they did a section that was a little deeper on web server optimization and showing the whole full stack piece. So it was kind of cool. It was a good mix if you were more technical. They probably like their section better if you're more rummy, business-focused.

I don't know. With these thick JavaScript client frameworks, people really do need to understand RUM and the client and everything because, you know, if it's nine seconds to the client, the client doesn't care if it's eight seconds inside the browser. Yep. There's just one in the back end. Yeah, I mean, I think it was interesting because both of those areas, and Doug made the case on the web server side, he had a customer quote, which of course he didn't name who it was,

but said, you know, the web server should just work. I don't care about that. It's the back end. And he had actually gone through with them and found there was this huge delay because of an Apache config issue. And so it's an overlooked, I think,

area with our customers. And I agree with you on the RUM side, too. I think people turn it on, it turns on so easily, and then it's an overlooked i think area with our customers and i agree with you on the rum side too it's i think people turn it on it turns on so easily and then it's like oh set it and forget it it's like it's but do i actually know how to do anything with it am i actually you know using it to make improvements so yeah it was a good really a good hot day i mean fun ish i mean is eight hours of hot day fun now but i mean it was it was was a good topic. Just so people know some of your background, you came originally over from Keynote. Keynote, yes.

He's now doing a lot with our Insights team, where they're going to help you consume and understand that ROM data. We had you on the show before. I looked at one of your presentations. Really, really amazing stuff that you do.

The one with Andy was not as great, but no, I'm just kidding. Forgive that. That's fine. But a lot of really awesome stuff. You know, diving into all this data that you're getting from the browser performance, how people are using the site, how that can help you change your priorities for where to optimize your performance on the front end. Really, really cool stuff.

So I just wanted to give you a shout out for that. We've come a long ways from just a Harfile. You know, just a little. This is amazing. If you think of what you get, you do get water. You get the kind of basics of water falling in RUM, you can see. But then you can do so much more in the correlation, right? Well, and the fact that now we're enriching RUM with, like, user behavioral and business metrics.

Again, I think that's where, you know where the other session I was in, the power dashboarding, whatever it was called. Power dashboard. People love dashboards, so standing room only. I take no credit for that. Maybe John Kelly should take credit for it. People want to not just understand the performance of their users

but segment those users and understand how that affects revenue and conversion. It's an interesting place I think we're at where we're seeing this intersection of what would be typically business analytics tools like Google Analytics and Omniture and then performance, right? And we've talked a lot about that. I've been doing it for a long time like you guys have about, hey, performance matters to customers or to the business, but we've never had really a way of showing it as strongly as we do now,

which I think is pretty important. Yeah, let's face it. Response time equals conversion. Faster gives you more revenue. Very proven. Yeah, I think people are starting to realize, like, I'll get more funding from my business partners if I can show them that what I'm doing as a performance person or performance engineer actually affects the bottom line.

And I think it's super important. And, you know, we've kind of been in this little ghetto as performance people. And they're like, hey, business people, pay attention to us. But now if we can bring the data together, it becomes a much more compelling discussion. Yeah, absolutely. What is the biggest challenge that you see engineers have getting into dashboarding, right? Because sometimes we're going from logic to something visual.

What do you see that challenge being for them? I mean, John was really interesting in his kind of view of dashboards, and he did the first half of the workshop that he and I did together. And it was basically like, you know, dashboards are really these bright shiny things that people want, and they spend a lot of time building them, and then they tend

to not look at them ever. It's more fun to build. It's like Dungeons dungeons and dragons i used to love building my character and then when we actually did i was like yeah it's not that interesting yeah so that was his point is like and then you know when there's a problem maybe people do start looking at them right but his vision of the future was sort of this merging of dashboards and in the example he gave like our problem tickets which is really an internal web page kind of thing, right? Yeah. Where, you know, don't have a separate dashboard here and then a problem detection here, put it all together.

And he did some kind of cool stuff with that. That was experimental with some plug-in that I'm sure he got off the black market or something. I don't know. I don't know what he did from some other SC. You know, it's dark web. Dark web, yeah. I think John Kelly's on the dark web quite a lot.

100%. Like, yeah, 100% he's on's dark web. Dark web, yeah. I think John Kelly's on the dark web quite a lot. 100%. Yeah, 100% he's on the dark web. But I thought that was a cool concept of it's really dashboards were intended to give us information that we could consume easily, right? Update, see. And maybe we need to shift and think, well, maybe that's not the only way to dashboard data. It could actually be in other formats, I guess. So that was a cool session with his portion. And then I kind of talked a little bit more about

how we're taking the data out of Dynatrace and then mining it for more deeper user analytics, which is in the Insight practice. Yeah, it was cool. That does sound cool. I think of the old, I think John Ospaugh, way back in the DevOps world was, or web ops even in that world, was if it moves, graph it. And that was like the basic thing to say, all right, if I'm usually building dashboards for someone else, my stakeholder, who's interested in showing the value of what I write, the evidence of my value, my work as a team or as an engineer, build some dashboards for yourself. And if you're only having, it'll actually, wow, I keep track of a lot of numbers that

I don't, I'm not aware of. And you start putting them out in front of you and saying, you know, there's this stuff that I can just feed myself in a continuous way. It's like, okay. And then someone else goes, well, how did you know that? So, oh, well, I wrote this little, I got my own little dashboard. And they're like, could you share that with the rest of the team?

But it's a great way to get started where you're not having to be in front of an executive with pretty colors or whatever. You're just kind of learning about your own numbers, your own things that move, that make you go. You're kind of training yourself as a performance analyst using your own data and kind of using

Or even as a PM. It's like, I kind of take this data here and that using your own data. Or even as a PM. It's like I kind of take this data here and that data there. I was going to say you also got to be careful, though, about just throwing too much data on the glass, as we were talking about yesterday. You're at risk of just being data on glass. But that comes in when you start putting these things together. You start learning how everything works, and then you know what to piece together. And eventually you mature into putting the right things on the glass so that you do have that actionable data.

But you've got to start somewhere. But don't you think you have to have, and maybe because I come at it from a more business perspective, a lot of our inside customers kind of straddle that line between business and technology. But I think if you have a dashboard and it doesn't have some business metrics on it, which I think I see a lot of them that don't. Even if you're deep in DevOps, and I think it's still, to have technology specialists like us think about business outcomes, I think it's really a good thing. And so I think John was trying to make that point too, is make sure you are including the business impact at some level.

Well, I think that you bring up a really good point about DevOps. There seems to be a break in many DevOps organizations where we will compensate for speed, but not necessarily business impact. And, in fact, some of the development organizations want to specifically exclude any metric outside of what they can control for how they're compensated. So if their child grows up to be a criminal and kills people in production, I mean, not actual people. They're like, that's not my kid. I turned 18. I'm done. I built the code. That's not how I programmed it.

I didn't write that code. You're compensating me for eight pushes a day. Right, right, right. Not for the conversion rate or engagement. Not for having defects show up. We'll just fix that in the next build. Or increase cost. There's no cost impact to that.

That's interesting. And I always, to that point, like every dashboard I've ever created, I've always had a tie-in to the other side. So if I was doing like an infrastructure dashboard, I would have a, here's my either a small world map or just a single green or a single button encompassing all my user actions so if any one of them turned red you know i'm looking at my infrastructure all my everything's green on the front end great you know or vice versa if i have a front end type dashboard have a couple of notifications on the back end because oftentimes when you have an issue you'll see the back end fall over first before you see it translate to the front. So at least you catch that.

I mean, obviously you'll use alerting and all, but having that context front to back is really, really important. Just even one little drop of something there. Okay, here's my trifecta challenge. Combine dashboarding, ROM, and session replay. So what do you mean? Just think about those three. What comes out of that?

That's like a power trifecta. Yeah, I think it is because if you think about, I have a dashboard that has real user stuff on it, and I'm actually watching whether it's AppDex or it's JavaScript errors or it's abandonment, conversion issues, engagement issues. If I can then drill down with an in-context filter to the sessions

and then I can drill further into, hey, which ones have session replay? I think that's the power of session replay because I think more and more about it. It's like, if I have a million sessions a month, that's a lot of sessions. Having just a million sessions a month is not... It's cool, but it's how I bring it into context. But understanding session replay off of an iPhone today where conversion rate has dropped. Exactly. That becomes magic, right? That becomes,

and I mean, we have lots of examples of this, but we've heard, you know, working with executives where they'll see a session and they just get so excited. You might've spent hours and hours and hours building out a data model and like showing them all this great, amazing statistical things about, you know, thousands of users. And then you're like, here's one session and they're, ah, wow, let's get the team going. Let's fix it. You know, and I think there's a power in that too, right? Obviously in context with the right session, but it's a power in sort of seeing what's happening, which I think changes. And that's why everybody's loving coming over to the tower

and looking at session replay, even on easy travel. It's a fake website, guys, but they're getting excited because I can see how that would work in my environment. Yeah. Okay, then the fourth thing would be the query language, but that's kind of... No, but that's a really huge one. Actually, the fifth thing...

Whoa! What's the fifth thing? Penta, not trifecta. Trifecta, quadrifecta. But the query language is definitely a huge piece of it, right? Because, yes, you could, and I think you've probably been doing a lot in the past with exporting the data via the API and massaging

it. But with the query language, well, number one, the query language gives you the ability to pull what you want out into anything else, but you can also do a lot with the dashboarding within Dynatrace itself with the query language. So it allows you to do a lot natively right in the same space you're already working. Yeah, that's exciting.

I mean, it's a really great package. I mean, so this, going back, we were looking at some SaaS-only type solutions, and I remember I reached out to you about this one, but, you know, sometimes you can't put an agent on the back end because it's a SaaS platform. Yeah, you can't touch it. And the reason why I've been dying for Replay come out, because, I mean, if you're in one of those SaaS-only platforms where you really don't have much you can monitor as that SaaS customer, if you could do the real user monitoring, the session replay, the synthetics.

Don't forget about synthetics. It's my history, man. Own the entire front end. That's a really, really powerful package. That's going to at least get you to know exactly everything that you can control. And then if there's a little black hole called your third-party support, this is

on you guys. I was on a call upstairs before lunch, and I won't say who it was with, but a very large retailer. And that's exactly what they're going to start with, is they have all these internal apps that are used by their employees in their physical stores, and they don't have visibility into the end-user experience. So it's exactly what they're going to start with is they have all these internal apps that are used by their employees in their physical stores and they don't have visibility into the end user experience. And so it's like, they were talking at one point about, oh, we're going to set up Selenium synthetics in like little in the stores. And I'm like, what? That would be a

nightmare. But with Rum, you know, and that's where they're at now, even though they've looked at the whole full stack, it's exactly that. They're going to get huge visibility into, you know, tens of thousands of their employees and then dive deeper from there. Right. Which I, it's exactly that. They're going to get huge visibility into tens of thousands of their employees and then dive deeper from there. I think it's really cool, even if all you have is ROM. Obviously, we want everybody to have the full stack. In some cases, you just can't. Exactly. What is new in

the services part of your team? I think what we're trying to think about is we started Insights around ROM. Insights has existed for a long time. It really was built up around synthetics initially. And the whole concept was don't just give people data, give them answers, right? We also say that in our product now.

That's pretty cool. Where did you get that from? Yeah, I know. But we're doing more and more of that with RUM data. And I think we're trying to bridge that business technology gap. So I think Insights is kind of the sweet spot of I think Dynatrace does a really good job of problem detection, AI, all the stuff we talked about at the conference, where it still doesn't do a great job. And hopefully, and hopefully no one's listening that's going to fire me. Constructive feedback. Is, you know, it doesn't do a good job of sort of the, you know, month to month trends, summaries,

10,000 foot view, kind of zooming out, right? And predicting a little bit where you're headed. And finding hotspots that maybe are not, like, problems, like, that are going to trigger a problem detection, but, like, things are creeping up or, you know. Keep an eye on it, bucket. Exactly. Everyone forgets to keep an eye on. Yes. That is a cost space.

Yeah, that's a cost space. Yeah. So, I mean, that's why I think. Sorry. I don't know why I just touched your shoulder. Yeah, that's okay. People do that all the time. It's Las Vegas. Is it. Yeah, that's a cost space. Yeah, so, I mean, that's why I think – sorry, I don't know why I just touched your shoulder. Yeah, that's okay. People do that all the time. It's Las Vegas.

Is it? Yeah, okay. And it's nice for podcasting. We don't have an HR department. Yeah, okay, cool. Excellent. I can do whatever I want. I should have a beer right now.

Where's my beer? A bacon-wrapped beer. Yes, bacon-wrapped. Kroger. Shout out to Kroger. But I don't know where I'm going. Oh, so what is Insights doing? I mean, I think what we're trying to do is people are paying and getting all this great Dynatrace-rich data,

but let's understand that there's different use cases or different owners that can consume it, right? So your executive is not going to be in Dynatrace necessarily doing peer pass and even looking at RUM data, but they might be wanting a monthly analysis that looks at, of the people that are dropping out of the site, why is it performance related? Then I can get my team to focus on that. And so that's the sort of insight that we're trying to bring to the product or in the services area to our customers is there's different consumers of the data. The data is super useful across the technology and business spectrum. If you can get it out in a way that they can consume, it becomes really powerful.

And then we can give them expertise kind of about, hey, we're working with hundreds of other customers. Here's how they're optimizing their CDN. Here's what they're looking at in terms of JavaScript frameworks. Here's what they're doing around image compression with Akamai or whatever. Right, right. Yeah. And not to sound like a commercial for us, because I work for Dynatrace, but what I really

like about that. You work for Dynatrace? I do. But what I really like about that. Yeah. What did you say? Yeah, you're going to have to pay for your fighters. No, but the fact of the matter is every time, you know, I used to spend a lot of time, a lot more time with customers on the sales engineering side.

And we'd be showing them all this data, especially with the Atman, where it was always a little bit more complicated. The Atmans, we should call them. The Atmans. The app mons, we should call them now. The app mons. And the biggest issue customers a lot of times have when they have these tools is, well, this is great, but we don't have anybody, number one, who has the expertise to look at this stuff or maybe has the time to do it. So the fact that we have services. And how to make it actionable. Yeah. Right.

And that's us. So it's like, hey, you hire people for everything else. We have services. We have the experts. I mean, you should see some of the stuff Ben's done. It's really awesome. But I love that we don't just, here you go, do it. We're like, hey, let's build a service around that.

But I like the trending and prediction is also a really important thing that was harder to do in the Atman world. I don't want to say impossible because you can't do just about anything in there. But it was immature. But I think, again, connecting the technology to something meaningful in the business, or if you are an executive or a manager over these things, you're like, show me excess capacity. And we just talked with the Pivotal guys about min and max kinds of things. They're like, all right, am I paying for excess capacity? Can I not pay for that? And do I need more elasticity?

So there's some bigger, longer-term questions that they don't have data to answer in a big, long-term way. So, wow, we've been running for six months over capacity. Why? What do we need? You know, I can't see it. Now you can see it. I ran into a customer recently where a third of their server capacity was taken up with 500 errors. They were unhandled from

bad requests from development code. So either wasted cycles, over-provisioning, or they may know strategically what's happening with the business in a number of months, and they never pick up the phone and call and say, by the way, you should buy more machines, or you should

optimize your code. So now you can see things coming in the future that says, hey, here's your target. And get on the trajectory to get into that. So yeah, insights is going. Anything else? Yeah, I mean, I think we're thinking about how does Session Replay fit in, right?

Because, you know, I mean, we're building this kind of ROM analytics engine, which our team, you know, got to play around with Spark and Databricks and Tableau and a bunch of stuff, which is awesome. But now the question is, okay, if I have this great RUM data that we're sharing with executives or business people or technology leaders, too, how do we session replay in that? Can we actually take some of that raw data and do something that maybe is not available in the product yet, like, you know, heat maps, right, where you can see where people are clicking. Can we use the RUM data to say, here's a cohort of users that are struggling, and then I can filter directly back into the product? So we're kind of thinking about, with all the Diantrace-grade APIs in and out, what can we do to sort of make that even clearer to a business person?

Not just a single session, but how do I actually use the data to drive into the right sessions? Oh, yeah. And so, yeah, I think that's where we'll spend probably the first half of our next fiscal year is focused on that. Yeah, just kind of innovating around that area. Oh, that's exciting stuff, man. I can't wait to see that. That's going to be great. Did you make any performance resolutions for the new year?

Oh, yeah. Performance resolutions. Yeah. Huh. I mean, like computer performance. Like Brian and I have to do 50 push-ups today because we had pizza last night. And so that jives with my resolution. Yeah, no.

I don't think I have any good resolutions for this year. Maybe drink a little less. Not get fired for saying stuff on the podcast. Actually, that was my wife would say. I was like, make sure you still have a job. So that probably is my number one resolution, actually, is don't push my humor or my swearing too far, but I get fired.

So yeah, that's a good one. There you go. That's something I'm always on the lookout when I'm on the mic. Like, what am I saying? Exactly. And how many performs have you been to? This is my fourth perform.

So you were in Orlando. I was in Orlando. That was our first podcast in Orlando. We had just gotten acquired, and we were like, who the heck are these Compu or Dynatrace people? I remember a good friend, Steve Filoni,

was there. Steve and I were HP people. Yep, yep, yep. Exactly, yeah. So we were part of the Lode Runner Performance Center world. And then I had jumped independent and Steve was at that Orlando thing.

And I'm like, we've got out. Yeah, awesome. That was cool. I will say I like Vegas quite a lot more than Orlando, but that's probably just because Steve was at that Orlando thing, and I'm like, we've got out. Yeah. Awesome. That was cool. I will say I like Vegas quite a lot more than Orlando, but that's probably just because I'm based in Phoenix and it's a 45-minute flight. Yeah.

Well, plus it's Vegas. A little selfish perspective. Yeah, I mean, yeah, exactly. Totally selfish. Well, yeah, it's not as humid either. No. Awesome. All right.

Cool. Thanks a lot for coming by. Thanks for having me. Thanks. Hope you enjoyed most of your time. It was great. Thanks, guys. Thanks.