PurePerformance - Dynatrace Perform 2017 Tuesday Morning

Episode Date: February 7, 2017

It's time for PerfBites. What the f*** is PerfBites? The fourth square meal of the day. Don't bogart the PerfBites. F*** waffles. Microwave ready. Add nutritional value to your brain. I'm Jackaloo.

It's time for PerfBites with your hosts Mark Tomlinson, James Pulley, and Howard Chorney. PerfBites. Whatever. Good morning, Mark. Good morning. Good morning, James. Good morning, Brian. Good morning. Good morning, Mark. Good morning. Good morning, James. Good morning, Brian. Good morning.

Good morning, Sprocket. We are live at Dynatrace Perform 2017. This is day one in the morning after a rather crazy welcome reception last night. You could say that again. Yeah? Yes. Where's Bob this morning? I saw Bob, and he looks very different when he's not wearing a hat.

Yeah? Yeah. He's a very different looking guy. I wonder how he is when you get some caffeine in him. I don't know. That's crazy. Well, this is sort of a special couple of days of broadcasting live. We're outside of the breakfast.

There are people walking by. How was the breakfast, by the way? It was good? Thumbs up to the breakfast. You guys, did you have eggs? Biscuit. Biscuit? I had a biscuit, but I also had to get a bagel because coming from Denver, moving from New Jersey, North Jersey, to Denver, bagels are just chains and chain bagels.

And there's nothing good, so I had to give them a shot. But I got an everything bagel with onion and garlic, so I'm hoping that wasn't a mistake to our guests. And I'm enjoying it vicariously right now. Yes, that's awesome. So one of the interesting things for our normal PerfBytes listeners, of which there are several people here who listen to PerfBytes, our co-host, our guest host co-host, we're simultaneously broadcasting through Pure Performance Podcast through Brian Wilson and Andy Grabner, who's gallivanting about being socialite.

We may see him eventually on the mic. We may see him. He's got a lot of stuff on his plate. He's one of the brand ambassadors, so he's being pulled left and right. What do you have to do to become an ambassador? You have to sell enough product. Who's got the brand ambassadors? Yeah.

So what you're saying is he doesn't have to be appointed and go through a Senate confirmation? It's not like a cabinet visit. None of that at all, no. Okay. I think as long as you're coming from Austria, you could be an ambassador. Yeah. Oh, okay.

It's like an Arnold syndrome thing. Like, you know Arnold? Yeah. Then you can be an ambassador. Ah. Yeah. Harold last night was funny. Yeah.

I'll be back. But speaking of your listeners, what I wanted to mention is to our much smaller listener base is that we are also PerfBytes. If you don't know who they are, which is probably pretty rare, but there might be some people, they should check you guys out. Because you're one of the top performance podcasts. I think we were the first and only dedicated, exclusive podcast. And then we got you guys.

We had Andy on the show a couple times. He was on a live show with us. Yep. And then we started with you guys. So you can go to perfbytes.com slash live. Some of you might be listening there right now. And do you know of a URL? Did you get a URL for your player on the Dynatrace site or the Performance site?

Yeah, but sometime today it should be on the regular Pure Performance website. So you can go to dynatrace.com slash pureperformance. It's also, I don't know the URL for Perform 2017, but if you go to the Perform 2017 site, we are one of the tabs on the page. It all loads simultaneously. We'll look that up. And, you know, this is the second time we did this. I have to tell you, James, I have TweetDeck installed.

This is TweetDeck on a totally different machine from 2015 when we were at Perform in Orlando. Yes. And it's a totally wiped machine, installed TweetDeck, logged in as PerfBytes, and guess what the columns were? It had remembered all Dynatrace Perform 2015 hashtag in TweetDeck. So you can tweet things now, and I didn't have to set anything up. It was awesome.

So what you're saying is that your application cyber-stocked you from machine to machine? That's right. All that memory was in the cloud somewhere. The TweetDeck was tracked, and it was performed 2015. Not that we haven't tweeted anything since 2015, because we have. Yeah. Yeah, exactly.

You know, that does bring up an interesting story that I read this morning. Really? Somewhat performance-related or metrics-related. It was Vizio. Oh, Vizio. Vizio had to settleizio Vizio got had to settle for spying on all their television customers because they were

at least anonymously which is good, right? Yeah collecting second by second information on everything they were watching on TV and everything they were clicking on TV Really? Anonymized data

but for all their metrics So like Nielsen ratings type competitor I guess I don't think they were doing it for Nielsen ratings type competitor, I guess. I don't think they were doing it for Nielsen. I think they were doing it more for user behavior and other things like that or maybe looking to sell that information. Who knows? But I think they got settled for $2.2 or $22 million, one of those two.

It was kind of a small amount if you consider the class action. Just say no to spying. Yes, but that was your tweet deck was spying on you then in that same kind of way. Well, but I clicked on the terms and conditions when I signed up. Right, they didn't have terms and conditions. And they said, now we know. The statement was, now we know it's good to get people to agree first. Yes, kind of like the iTunes terms and conditions.

There's things in there about your firstborn child, donating your kidneys. It's all in there, right? And we won't even go into the South Park episode. All right, don't. Yeah. donating your kidneys. It's all in there, right? And we won't even go into the South Park episode. All right, don't. So, James, you want to kind of kick us off with a context for why we're here? So I think there's no better way to discuss why we're here at Perform 2017, kind of the preeminent conference on performance at this point, is to take a look at a failure. and kind of the preeminent conference on performance at this point,

is to take a look at a failure. And let's take a look at a recent one. Let's take a look at 84 Lumber. We talked about them last night. We did speak about them last night. And we're going to speak about them just a little bit more because we want to do our standard kind of deep dive on 84 Lumber and see what they're doing. The back story is they're on i-84 somewhere um we assume yeah but that's where the and these guys

talked about like redneck people getting on that getting on i-84 lumber and like it's not lowes it's not whatever but like people never use the web it was kind of funny last night yeah so so 84 lumber bought a commercial on uh the super bowl which i understand is about five million dollars for every 30 seconds they bought a 90 second commercial so let's assume they spent somewhere in the neighborhood of 10 to 15 million dollars for a commercial um then this of course drove traffic to their website and and brian what happened to the website website kaput kaput boom goes website go boom yeah yeah exactly it's a it's the classic story of a super bowl ad it is the classic year after year after, there's something like this. Exactly. You'd think it wouldn't happen again. So you spend all this money on your Super Bowl ad.

Let's take a look and see what they did on their infrastructure side of the house. And so for that, we turned to one of our partners in our effort, which is GT Metrics, our deadbeat sponsor, to see what they have to actually say about 84 lumber. And in this case, we take a look, and they get some pretty bad scores for general practices. We don't have expiration headers. We're not managing our content in a CDN effectively. We're making a lot of requests. We have a lot of different DNS lookups that go on.

So essentially we've got very, very low scores. On one side of the house we have a D from Weissloh. We have a B from PageSpeed. But both of them agree that leveraging browser caching and CDN was not handled effectively. So here's my criticism of this for something like a Super Bowl launch. Browser caching may not make a difference if these are first-time visitors anyway. Ah, but CDN caching would. Of course.

And this particular site, 84 Lumber, is hosted in Microsoft Cloud Azure. So they were probably assuming that auto-scaling would work and it would handle all of their needs. Yep. Now, there is a CDN available inside of Cloud Azure as one of the options. So did they flip the switch? Because I think it's one of those sort of easy no-brainer CDNs, like give me CDN, click, please. Yeah. Which is the easiest thing you can do. NS Lookup apparently points to a

direct, referenceable source inside of Cloud Azure, which is not a CDN node. So sad. Just imagine all of those projects that are not going to get built now. No lumber, no supplies.

I mean, that's sad. There was a housing boom once in this country. You can build a house. You can't build a website. Yes. If you would have just taken... Now, hang on. They maybe don't know how to build a house. They just give the supplies.

Right. Hey, everybody. Could be. Right? And I think Andy's arrival is perfect time because the one thing I wanted to bring up in terms of this, right, and this is the kind of thing Andy was talking about, is this is your classic disconnect between marketing and the technology team. We've never heard of that, Brian.

Right. Now, this is 84 Lumber, whose Super Bowl ad went down. Right. Now, and especially, you know, DevOps is the big word these days, and everyone's trying to migrate to it, and that's all this idea of feedback. Where's the marketing department in DevOps? Why didn't the marketing department kind of hand off a million dollars to the engineering side of the house? But not even that. Did they say, hey, this is the traffic we're expecting to generate, do they give the development teams what they're expecting to be able to support so that the development teams can test

for that? I don't know. And give them an opportunity to test for that. What's going on? Lumber, they went down. I don't know. During the Super Bowl. Sorry, I came in late into the discussion. We were talking about this last night, too.

It was a spectacular. By the way, that's Andy. Oh, yeah. Yes, this is Andy of Pure Performance, ladies and gentlemen. Exactly. And actually, I have Harold with me as well. Good morning. Yeah, you guys survived well.

So, Davy Jones had gone through last night about some of the 84 lumber. James is bringing up some stuff, looking at it on Cloud Azure. It's a Cloud Azure app. And then we were just making jokes about how sad it is that the houses are not going to get built because no one can get the lumber. But Andy is here to take it off the joke side and bring it back

to our track. I've been told that I'm really serious. Or too serious. I try to be funny, but actually it never works. It never works. It's funny. I know. It's hard to have a podcast with two straight guys. As in stoic and humorous. I just want to clarify that. I know.

We had this discussion last night. What straight means, right? Yes. Classic disconnect. We've never heard of it before between marketing and engineering. Right. And what's so sad is that marketing may actually try to present this internally to management as a success. Oh, my God.

We were so successful. We drove so many people that we crashed the website. And we've seen this over the last couple of years. We're redefining failure as success. We had to come up with a term for that. I have a lot of terms, but I can't say them in public well andy and he's got a split again already yeah unfortunately well we actually start so for those folks that are listening online that actually at the conference get over to we're getting over to

the um main stage yeah yeah so everything's going i trust you guys to do a good job here because i obviously don't and i'm not funny i cannot add anything that is hilarious to the mix. Two things we'll say. One, Andy, if you will go out and find people and bring them here to tell a performance story or a testimonial. Secondarily, we will collect comments about you in real time

so when you come back, people will say things about you and we'll share that with you. Wow. Instant is instant feedback. So if you have an idea or an opinion or a comment for Andy, you can tweet it to, it's pier underscore DT. Yeah, at pier underscore DT. At pier skid or at perfbites. All right. Sounds good.

It's a challenge. And I'll try to write you some material. Because if I make up a joke, you know it's going to be good. All right. Of course. All right. See you later. All right. I'll see you over there too. So folks are headed off to the main stage. know it's going to be good. All right. All right. See you later. All right.

I'll see you over there, too. So folks are headed off to the main stage. Yeah. I should probably be going there. Do we even know what the main stage presentation is? It's your CEO, right? John Vincent. I'm not sure he's doing it.

Let's pull up the app. And if you don't have the app, everybody. Yeah. Oh, there is a Dynatrace app. That's true. Yeah, for those of you over here. Hello, sir. How was breakfast?

It was good. Was it good? What did you have? I had a pineapple. Pineapple? Yeah. I had some pineapple, too. It's a great way to start the day.

Some fresh fruit. Very nice. Hi. How was breakfast? Excellent. Excellent. Did you have proteins or fruits? Everything.

Had everything. They just got an everything bagel? It's welcome to perform 2017. Yes, Mr. John Van Sicklen will be doing the speaking. Probably talking about, I don't know, the business. Dynatrace. Giving the business. Giving a business update.

Yes. We don't really. We're geeky engineers. We don't really do stuff like that. I think what it'll be doing probably is be hinting at a lot of the unveilings that are going to be going on today. Are there some? Updates on a lot of the new things and roadmaps and what's just around the corner.

He'll probably tease those. There are some sessions later on that are going to cover some of this stuff a little more deeply. I'll sneak away. My session is at 10.15. I don't know where. I've got to look that up. It's all in the app. session is at 10.15. I don't know where. I got to look that up. It's all in the app. It is all in the app?

It is all in the app. Absolutely. You know, I do have to ask, what did you have for breakfast, though? Because you're asking everybody else. I didn't have breakfast. Oh, I'm sorry. I wasn't feeling the package. So I sort of just let it slide, so to speak.

Okay. Yeah. No big deal. Anyway, so you have your 10.15. And then are you doing any other slots? I'm done. All I have to do

is podcasting for the rest of the day. Okay. And, you know, I probably have my real job to do. Well,

I myself am going to try to get to some of these sessions because it's kind of my job. And then come back and report to us. Well, yeah, I'll be very happy

to talk about them. All right. This will be good. And I remember in 2015 when we did this, it was small talk until everyone had been to sessions. And so in the next section, it'll ask people what session. There was excitement. And then there was the late night party like last night that things get a little raucous.

It was kind of fun. So with that, We'll see you guys Let's see It's a little after 8 o'clock And we'll see you in See you on the flip side As they would say Alright

Any closing remarks? How do you feel Being at your second Dino Trace Perform James? It's cool. So articulate. And I do want to issue you a challenge, James. Since we're on a trend, twice begins a trend. Actually, what defines a trend?

We'll get to that later. So two points define a line. Right. And so I think you need three to get a trend. Let's turn this line into a trend. My challenge to you is every time we get on the mic, I want you to figure out somehow to bring up 84 Lumber again.

Well, new information. Right. New information. Can we get someone to reply to us from 84 Lumber? Let's barrage them on LinkedIn. Let's see if we can find someone on twitter who wants to join us all right how would you do oh yeah yeah computers because i was gonna say i didn't see a i don't see like a handheld telephone here wired to the wall yeah exactly so bring them in on

this is being broadcast on a pots line right now it's a 144 modem. Plain old telephone system. Isn't that the POTS line? Never heard of that before. I learned something new today. That was before you... I learned something new about Dynatrace yesterday that I did not know existed in the getting started. I was like, you could do that?

I want to share it because a question we often get is how much data of the continuous, because we're a continuous rolling window of data for the deep dive data, depending on how much disk space you give, like two terabytes. And if you're a heavy traffic site, that might give you 10 to 12 days of rolling data. Right, so there's aging and that kind of stuff. Right, right, first in, first out.

And the question is always, how far back does my data go for this app? And it was always a question, and we have to go into debug mode and look up the files with time. Yeah, yeah. Well, no, this isn't a roll-ups. That's the performance. That's the trending data. This is like the deep, deep data. Okay.

But now. Performance warehouse data. It's just database. Right. Right. So, and in the past, you'd have to go into debug mode, look at the session files, and look at timestamps and try to figure it all out. Now, and I don't even know when they introduced this, you right-click on your system profile, look at details, and it tells you right then and there what your oldest data point is. I was like, oh, my gosh.

So anybody who's curious. That is in the name of transparency. Yes. I was quite amazed. And it doesn't do a live update while you're looking at it, though, does it? No, no, no. That would be kind of awkward. Probably periodic. Yes. I was quite amazed. And it doesn't do a live update while you're looking at it, though, does it? That can be kind of awkward. Probably periodic. I will say having been a PM of

a product, you do sit down periodically with the customer support people and say, what's the list of things that you have for cases? What are the call frequencies and the topics? And it's how you parse that, not just quantitatively. Someone will say, what are the call frequencies and the topics. And it's how you parse that, not just quantitatively. Someone will say, what are the things that keep coming in that are low-hanging fruit?

Like the call duration is really short, and it's an answer that you're just tired of handing out. So in the old days, we just put it on the FAQ. Right, right, right. Then we learned how to take the FAQ and put it into tooltip suggestions in the help in the app. So you can pull the service that pulls the FAQ and put it into the app. Now you just build it into the app. It's not a tooltip. You actually create a feature.

Yeah. Or one of the things I know we were going down the line was I think we were looking to put Clippy and the little dog. Oh, no, no, please. Wasn't there a walrus one, like a little purple? Don't you remember Microsoft Bob? I never got Bob. I never got to use it, but I love referencing it. It's one of my favorite references.

There was a little walrus thing. Wasn't that the 98 themes it turned into? Was the walrus part of the theme packs, though? No, I think it was a separate. Remember how heavy those themes are? Oh, yeah. It would take up like 25% of your CPU just to animate the little guy, Java applet or something.

That was stupid. Well, I don't remember the walrus. I remember the dog, and I remember Clippy. But there are people in the world who are like, that's why I bought a CPU. That's what I'm going to use my CPU to do. And now we're becoming energy conscious, so you're like, I'm burning a lot of energy here, a lot of watts, just trying to get the little animation guy to do his thing.

It doesn't really work on the radio when I'm animating myself. I was going to describe your dancing. I was happy the last time I went to Redmond to dance on Clippy's grave. Exactly right. Do they have a marker time I went to Redmond to dance on Clippy's grave. Exactly right. Do they have a marker? I went out there. I danced a little jig. They should have served money.

Poured a little bit of a pine on the ground. Buried a paper clip and gave it a little memorial spot. Like a real biker. By the way, we should also announce to the listeners of PerfBytes who know James and some of our new listeners in peer performance that James, he's bought himself kind of a cap and he's kind of letting the

manly hair come out, so you're going to be Hipster James. Is it a chapeau? No, it's sort of a character, maybe an alter ego? Yeah, Hipster James. You would just adopt this persona every now and then. What brought you to sort of go down this path? Any life change or something like that?

No, no, just... The joy of experimenting with your appearance. Sure. I just got lazy on... I've been shaving the entire noggin for the last five years. I just let it come completely out. You look like Gallagher. I decided to. He had a hard time on stage.

That would be frightening. And so I just decided, I'm a little bored getting up and applying the razor every morning when I haven't had anything to eat or drink yet. Right, yeah. So I said, well, Wahl makes a great razor. I can cut it on, and it'll go buzz. And you also thought Monty Burns doesn't quite make such a great role model. Well, Monty Burns has the horns of management.

Oh, that's right. But he does have it nice and waxed. He does have it nice and waxed. He does. And it was time for a new random orbital. So it was either let it grow or get a new random orbital for the noggin. I'm seeing, just sorry, I'm observing a lot of people walking back and forth to and fro. I believe that's the way you say that.

From like the breakfast eating area and then over to the main stage. Did you guys, when you registered, do you get t-shirts? Are there t-shirts this year? I'm seeing a lot of people with the same Dynatrace t-shirt. Not employees, but attendees. There's the Dynatrace store. There's all sorts of good stuff. So if you use your app and you collect points, you can trade it.

Yeah, I got that on my back, too. You got a Sharpie? I could just do a Sharpie and get rid of that. You could just put the sticker. Take the sticker and put it over it. Oh, yeah, perfect. It's the right size. All right.

All right, so anyway, yeah, just to that point. So if you use, like, do things on the app and the app and post things and reply and whatever, you get points. And then you can trade your points in for some swag. And they're actually even giving out dots. Dot candy. No, the Echo Dots. Not giving out, but you have to get 100 points. And it's not too difficult to get 100 points if you post and reply and like.

Just use it like Facebook. All right. So we're going to get on the app because post and reply and like. Just use it like Facebook. All right. So we're going to get on the app because we have stuff we can chat about. Yeah, completely. All right. We'll send out the URL. We're going to do some research on the URLs for your live player. We've got our live player.

And we'll start pushing this out there. And if you guys have any questions about what's happening at Dynatrace Perform 2017, just tweet us. Send us an email. If you know my phone number, you can text me. Stop by our broadcast location. We are just outside the cafe. We are right next to the selfie photo booth. The selfie photo booth.

And if you're not here for Perform, we're here until Wednesday, so you still have time to fly in. Yeah. And come visit us in person. Do a same-day registration, get in on some cool stuff. We've got some sessions. Or if you are with 84 Lumber, please, you know, get on a plane, come on over, sit down with us, and chat about what happened. Yeah. You'll probably get a lot of exceptional free advice from everyone attending.

I'm sure they're getting a lot of free advice. That's a good point. Yeah. So there are some interesting sessions, folks that could hear this broadcast, and lo and behold, someone says, I love Vegas, and why not go? I'm going to make it happen. They probably wouldn't get here until Wednesday, right? Today's Tuesday?

Yeah. Well, some people can get here tonight. Well, first of all, I'll say Wednesday, 10-15, our good friend Bob Stoltzman is doing the automated Dynatrace deployments on CenturyLink Cloud. And if his presentation style is anything like his podcasting style, that's going to be an exciting, crazy thing. BT is here. I guess that's British Tele exciting, crazy thing. BT is here.

I guess that's British Telecom, I guess. Very good guess. They're doing some app performance stuff and monitoring in a hybrid cloud with DevOps should be good. There's an OpenShift introductory demo. Wow, that would be cool. I'd kind of like to

see that. And you can get to the Red Hat stuff. There's also some things that are not straight-up technical that I'm seeing here that are like the performance engineer career path development stuff. So we have a lot of new, like you say, there's some sessions for new people who may be new to being, last night at the welcome reception, there was that young lady who had two months working with Dynatrace. And actually some of the guys from my puzzler that were

here from Michigan, they were mentioning that we have one senior engineer that's been doing it old school, but everyone else is like one year or less and they have a guy from their team here, one month, brand new to Dynatrace. Yeah, I just ran into a guy there from Dave's Bridal who's six months in on it.

They're like they want to really ramp it up more. I do like this one, which is Len Nielsen's Whose Fault Is It Anyway? Fault main isolation with Dynatrace. Get it? Whose fault is it? Whose line is it? Total play on words. That guy's got to be clever.

I mean, just stand back. And you know, one thing I wanted to point out, which you may notice as you're scrolling through, a lot of these sessions seem to repeat at different times. Because when I was first going through it, I'm like, oh, but these are all at the same time, all these ones I want to go to. But then you go through and like, oh, they're doing it again later or tomorrow or something

else. So make sure you, before you throw up your hands in despair to say I couldn't go to that one session, it might be repeating. Right. Now the other one, there's some stuff about HTTP2. So if you're building an HTTP2 app, and we know load tester himself, Scott Moore, does a lot of work

performance in HTTP2. You can hit him up. Oh, there's also some IoT type items going on. Yeah, there was a workshop on IoT you guys were talking about last night. Yeah, yeah, yeah. Who's doing that? Who's doing that one? I don't know. Zurich Insurance. Was that Lewis?

What's the one Lewis is doing? Lewis is presenting something. He's also doing something IoT, right? Algorithmic IT operations and machine learning from Moogsoft. I like those keyboards. No, I don't know that that's... It's probably not the same. Moog? Moogsoft.

Yeah. All right. So, yeah, there's some really, really... Oh, hey, DevOps for dinosaurs from the Razorfish guys. Didn't you know what that is? DevOps for dinosaurs. What are they doing? Like incorporating DevOps into Waterfall?

Well, you could do that. Call it DevFall or something. DevOps, WaterOps. WaterOps. I'm glad they put a little humor into that because on a serious note, you know, that's like the biggest... Yeah, yeah.

Yes, to get serious, we will pump you up. You know, that's just like a big transition, right? Yeah. And that's what scares everybody, changing and all that kind of stuff. I cover a little bit in my world. Yeah, we see it all the time. So I like that they're putting a little humor into it to maybe ease people in.

Because, you know, I work on this sales engineer side. Right. Coming from load testing and all that. And I got to tell you, if I was back on the other side, I'd probably be, I'd have had a lot of sleepless nights going through that. Yeah. Trying to figure out myself how to change over and also how to get the company to support making those changes to make it survivable. Yes. We're with you.

Also the idea of archaeology. So, you know, I still find people running an old pirated copy of Lode Runner 8. You know, that's dinosaurs, archaeology. You can find some bones buried somewhere. The old pirate bay before they shut down pirate bay. Yeah. Pirates, arg. Arr.

Exactly. You do a great pirate, by the way. Arg. Awesome. All right, so everyone's off. Are you going to go to the main stage? Yeah, I should probably have a presence there. Yeah, I'm going to check in on the world. We'll be back with you listeners in about, who knows when we'll be back.

A little later today. Momentarily. There'll be another show directly behind this one. Just like the trains. Just like the trains. So, yeah, there's another train directly behind this one. Just like the trains. Just like the trains. So, yeah, there's another train right behind this one. Perfect. All right, we'll catch you guys next time.

Bye. Bye. Bye. You know, and just by way of introductions, this is James Pulley of PerfBytes. I'm sitting here with Brian Wilson of Pure Performance Podcast. We are co-broadcasting from Dynatrace Perform 2017. Yes. Live from Las Vegas and the Cosmopolitan Hotel. Yes. Live from Las Vegas and the

Cosmopolitan Hotel. Yeah. Does this mean we've got to do our bad Elvis impersonations? No. Yeah, yeah. That's a pretty bad Elvis impersonation. Yeah. Let's hear your bad Elvis impersonation. I don't do Elvis. Just go for it.

Go for it. Then we can move on. Ladies and gentlemen, Elvis has left the building. That's as good as I'm going to get. All right. Now I see what happens when I try to pressure you. Yes. All right. Nothing.

All right. So we just had, I don't want to say, let me look up the proper title because I don't want to say it was the kickoff. I mean, I guess it was the kickoff, the opening speech. Yes. That just all happened. Let's see. The welcome, and we talked about the technology roadmap, and this is our future. This is always our future.

I know you had to hop out just a few minutes. I caught almost all of it. You had to hop out a few minutes early to make it back over here to our broadcasting booth. Yes. Where we put the broadcasters behind glass as like a zoo exhibit. Yep. What were your impressions from the kickoff? A lot of cool stuff.

There's really starting to become a convergence. And before I continue, just for the pure performance listeners, typically, even though we are Dynatrace employees who do the podcast, we try to keep it non-advertising. But obviously, just throwing it out there, we're at our Perform event. It's all going to be about Dynatrace this week, fellow people. Absolutely. So I'm going to talk about it unabashedly.

I'm not going to try to hide it at all. So some of the really great things is really just seeing the vision for the convergence of all the components in our suite. As you might know, might not because I don't know how much you follow our products, what originated as Ruxit, which is now just called Dynatrace or Dynatrace SaaS Manage, which was originally designed to tailor towards those large microservice, humongously deployed applications where traditional tools just can't survive in that world. What do we call that? We call that the Netflix model these days, I think. The Netflix, the eBay model, the Etsy model.

If you think about any APM tools, none of those are going to be able to scale. None of them are going to have the flexibility to go up, down, left, right, and also just scale with all that data. So what they did in 2014 was they started development on this new tool designed and architected specifically to reach this goal. That has been coming a long way. That was the initial goal. So in tandem, that's really really it's designed for the cloud since all the microservices is really the cloud

Ruxit come Dynatrace designed for the cloud or it's designed for the Netflix it was designed for the Netflix model as you say but it wasn't solely for that the idea was things are changing a lot. Things are going to have to scale.

And the first foray to doing that was to attack that model, but now where it's starting to go is now we're starting to converge data points and everything else all together with these. Because now that that model is mature, they're now working on bringing

all that together with everything else that we have already and making it a single product or single unified component that does everything. So that's really cool to see where that is in the roadmap. So let's talk about that integration. There are a number of ways that you can integrate. One, you can integrate visually. Right, that's where we started last year. You can have the same look and feel across the family. Correct. And some drill down capabilities from

one to the other, but yes. So the next way you can integrate is I have modular components that are shared across the family. So I may have a component that's in product X and product Y and product Z. And so we get that glue, not only look and feel, but functionality that crosses products. So are we looking at really both types of integration in this case? You know, I'm not 100% privy to all the things that are going to be going on. I think it might be a little bit more akin to a third model. So one of the big features of the newer platform that we have, just called Dynatrace, is this gigantic artificial intelligence engine behind it to do all the analysis of the data coming in. So we already

have the similar UIs and all that kind of stuff, right? So where I think where I'm pretty sure it's probably going, but again I can't speak with authority on it, is feeding data collected from some of these other components all back into that artificial intelligence for the data processing and analysis and bringing it back in. So that's been a big theme in Vegas this year. I don't know if you saw the Consumer Electronics Show earlier this year.

NVIDIA talked about their connected car platform and their intelligent car platform and really this artificial intelligence in the cloud. In the case of NVIDIA and their car platform, of course, was dealing with all of the soft processing that goes on as you drive a vehicle for how you identify objects and how you avoid objects and how you target cyclists and things of that nature. So let's talk about the AI backend for Dynatrace then. Correct. Give us some more details.

Well, again, it's not the product that I specialize in yet. I'm still trying to find the time to come up to speed. But one of the things that you see, which is always most impressive, right, is it's taking in all these feeds of data from the OS level, from the application container level, from the code execution level, because all those peer paths are still behind everything. It's looking at your cloud statistics. because you'll even see what the AI does is instead of you having to look at dashboards and wait to get an email of 50,000 alerts when there's a problem that goes on, it's taking all the data, it's going to crunch it all together, and when it comes up with an alert, it'll say there was a problem. We analyzed 37 million data points to be able to tell you your problem was X.

So then will it tell you go address Y to fix it? Or that's kind of the next generation downstream. Well, it would be go address X. Okay. It's going to be you had a severely long database query running. Okay. had a severely long database query running, or there was a service call that, with the new deployment, is executing 50 times per every transaction. It's just way too chatty.

It's telling you what that component is. But the really interesting thing that I just saw in the presentation, so Alois Reitbauer, who's with the company, he works on a lot of this stuff, and we have a chatbot part, Davis, right? So you can talk in natural language, say, hey, Davis, show me what happened last night. There's now the ability to, in different interfaces and different components,

and I'm not going to describe this well because I just saw it, but you can now say, hey, if a certain type of problem happens, take this action. So he demonstrated with the audience where he deployed a build. We all pulled it up on our phones and the build was like, you know, came up as a failed build. And as we refreshed, suddenly

the build was fine again. And what they were able to do in the product was say, if XYZ conditions are met, trigger a rollback or take some other action. So it's interesting because someone was just asking me the other day, a customer was like, hey, can you do remediation? And I was like, well, it's not there yet, but we have some old-fashioned ways

you could execute a remote command. This now is like full-on remediation. So I think it's getting to that point where there was a failure, it fixed itself, and by the time he got the alert, the problem was gone. So one thing that pairs very nicely with AI and performance is performance, in many cases, is really about patterns. Or in this case, what I like to term them as performance anti-patterns. If this pattern shows up, that means performance is poor in a particular area. And it may not even be reflected in, say, a particular

slow-performing database query. It could show up as evidence in other areas. But generally, overall, efficiency drops, performance is poor, and there's a way to identify it and mitigate against it. Yeah, I'm really interested to see how the AI develops, particularly if partners or other interested parties, hey, you know, I'm an independent as a resource. You know, can we add our own intelligence rule sets to the AI, the patterns we look for? So I think that's something that's very exciting that's going to be developing. I don't think that it's been thought out that far yet, but I want to see that capability. I will do a shameless self-plug and say if you get a chance, go ahead and sign up for the free trial and check out what we've got going on there. Because there are some really, really cool things going on there. Another good, you know, conceptually what it is, when we talked about this large-scale deployments, which really just get to be too big for anybody to really look at things the old-fashioned way,

is you mentioned, like, okay, let's say it's firing off too many database queries. Yep. Well, if you have 20,000 microservices running and you're running across three data centers, you're going to see maybe you get the alert first about those database queries because that's what you have something set up on so you think of a database issue. Meanwhile, it could be some small deployment that one team member made to one little service out there

that's making that service super chatty and being like maybe an N plus one problem on the service itself which is causing the databasing but the database isn't your problem, it's really that service. Or it could be something really esoteric. It could be, hey, on the microservice that was deployed,

somebody changed the log level on messages, and the logs are exploding. That's taking priority over queries. So it looks like a slow query to the database, but it's actually an OS configuration issue where those resources are tied up, writing logs to the local log. And AI would probably catch, hey, the logs are exploding in these three tiers across these services, and that's probably root cause for degradation of performance. Well, see, I would take it one further and say it's not going to know probably because we're looking at those transactions coming through and tracing from tier to tier to tier.

Yeah. That the AI is going to see the database queries exploding, but also trace back and see an explosion of those service calls or whatever else is going on there. And then it's going to know, okay, these are exploding because of this over here. Yeah, and I had a new build, and that's when this started. Right, right. So there's a lot of really cool stuff coming down that way. And then the other really cool announcement that I thought was out there was

IoT related. They just announced what they're calling the Open Agent, which will be an open source agent. So if you think of IoT, they might be putting an agent in a truck. Yes. And that agent's going to be there for 10 years.

So if you're thinking about developing a supported agent that runs for a specific technology or something, that's going to be insane to maintain, to keep up to date and everything. So I guess this idea, this open agent, is just going to be an open source blanket agent that you can fix up, drop in, and then forget about. So I can drop it on

an embedded layer that's going to be in place for the better part of a decade. The interface is supported for that long-term duration. And we can pull information over long-term trends. These IoT things aren't going to be being updated all the time, right?

And a lot of these... They're going to start hardening down these IoT things aren't going to be being updated all the time, right? You're not going to. And a lot of these. In fact, they're going to start hardening down these IoT devices. Oh, yeah. Based upon what's been happening on the distributed denial of service and compromised cameras and thermostats and everything. The last thing you want is for someone to get access to something on a truck, which then gets access to something else within the truck, which then causes it to lock the brakes while it's on the highway. Exactly. That'll be really cool to see that agent rolling out and the different uses for it.

A lot of exciting things coming down the pipe. I was plenty happy to see some of the stuff coming through. Hey, it's Lewis. Are you wearing your fat pants? Well, I was perhaps a year ago, maybe a little bit less now. But thank you for noticing.

I was curious, what did you think of any of the other items? I was impressed with the fix. Which one? The roadmap was very interesting. Where they're going with the other products and the integration they're doing with the open agent. Right. So are you all going to be doing some IoT stuff? Yes. Are you already doing it?

Yes, we haven't got a full clear path on it yet, but it's going to be one of the things we're talking about with our roadmap for budgeting throughout the next 24 months. You can put them in those pink Cadillacs, right? Oh, no. No, no, that's not my job. That's sales and marketing. Keep track of the sales and marketing.

Guys, keep up the great work. I've got to run. Not a problem. Thanks for coming by again. Thank you, sir, for your impressions. Hit bottom. I don't do impressions. We should actually...

Okay, we just heard an awesome Yoda. Yes. Alright, let's see if we can get somebody over. So, similarly, we've got Lewis pulling in, embedded agent, IoT, Davis, the artificial intelligence side of the house. So it seems like those are big themes that are showing up. Yeah, and I think IoT has been a buzzword for a while especially in the apm community but there hasn't been really much to deliver yet right a lot of people of all the

products right and i'm not bringing this just in general all the products i'd like to talk about iot i had time to look like they're on top of it but it's nice to now start seeing some things that are coming out in terms of apm monitoring so so let's draw back on the IoT front to give people a little bit of a primer. For those who are not familiar with IoT, you probably have some of these devices in your home already. They are the smart thermostat. Right. They are your Internet accessible alarm system. They are your smart television, your DVD player.

Your phone. Your phone, potentially. Technically, your phone, other embedded devices that are in your home, which could be in your refrigerator, could be in your AC system to notify you when to buy filters. And most major appliances, fridge, washing machine, dryer, they're going that way these days for self-diagnostics and reporting back. I was just about to say it's not necessarily about, hey, we have a nice fancy screen and we're going to tell you when your milk's going bad. It's also a lot of the big push behind it is from these manufacturing companies who want to know when a device is about to fail so that they can contact you and say, hey, we're getting a message that this component's going to fail.

We'd like to schedule a service so you don't have any downtime on your fridge, right? Exactly. So it has a large customer service component associated with it. Right. But the big push that we're seeing for IoT is more on the embedded side these days going forward. So it's less on the consumer device side. We're looking at manufacturing quality control, where we have intelligent

manufacturing robots, constant sampling of data. We have embedded sensors in the field, particularly for heat, cold, humidity, all sorts of environmental related items. You have seismographic data, which is now becoming connected. So it will instantaneously provide data throughout the day in case there's earthquake, explosion, sonic boom, whatever. But this explosion of IoT devices, this embedded technology connected to the Internet, is going to be hundreds of millions, is projected potentially billions of devices around the world. So you've got to be able to diagnose that back end to collect the data. Correct, correct, absolutely.

And also just even monitor the monitors. Yeah. And I think that's where a lot of the APM tools are coming into play, is saying, what's the health of that IoT data collection? Yeah. Is it getting back to its endpoints? Is it having delays? Is it having issues and all that?

It is not insignificant. Imagine if you have a sensor that trips when you hit freezing and the entire East Coast gets hit by a cold wave. Yeah. And then you have millions of sensors all of a sudden sending you data. You have to survive the swarm of data, so to speak, and you have to know that your analytics system can actually survive and provide you meaningful data as well.

Yep. It's crazy times. Yeah. It's an exciting time to be alive for a performance engineer. It really is. I'm going to have to go have a conversation with Davis, I think. You know, I always loved years ago, probably about six or seven, actually probably around seven or eight years ago,

mostly when the Apple phone was new. That might have been a bit longer ago. I think that was about 10 or 11 years ago. Yeah, my memory is working well these days. but what i love talking about exciting times for performance was when those sites would go down yeah it always because you know i was with the load testing crew and all and it was like that's right they didn't do their job right and people are now talking about performance and it was actually making the news about the site couldn't handle the traffic that was going to it and i'm like to my whole family that's what i do first of all right and i hope to avoid that problem yeah ideally i hope to avoid that problem yeah so exciting times indeed performances

been slowly rolling in it's at now it seems to be one of the focal points of all the, you know, not just the development teams and the operations teams, but the businesses. Yes. It's really coming into its own. You know, we do have a problem on the marketing front still, where marketing wants to redefine failure as success. Hence? Hence 84 Lumber. There we go. We got one in this time. Yeah, yeah. Advertising Age. Adage.com. Go and read the story online about 84 Lumber and the problem they had with their Super Bowl ad. Yes.

We kind of beat them to death. I forgot about it. I wanted to get you to mention that every time. Yes, yes. So, would you like to come on and tell us where you were in the morning meeting session? Could I trouble you to... So we have

Steve Clark with us. And you can either just get right on top of the mic there, or you can hand-hold it if you want. And Steve, can we say who you're with? State of Michigan. State of Michigan. State of Michigan. State of Michigan.

I'm the leader manager for the enterprise monitoring team. My team is responsible for all of the application monitoring that's done for all state agencies. Wow. Fantastic. So we've actually seen issues with state agencies over the last couple of years as more people are internet connected and you have these potential surges of activity at certain times of year. For instance, reporting on chemicals and storage to the state environmental protection agency in the state of Michigan, as an example, or tax time. And tell us how the cloud is helping to

address for the state of Michigan this issue of we have prolonged idle periods and then we have periods of total insanity for you know for particular agencies okay well the state has not adopted the cloud for those situations which is kind of sad so what one of our areas that would fit that is department of treasury and online taxes as you know we're right in the middle of tax season now. Everything's ramped up. After April, things will fall back, but we still maintain the same infrastructure, the same number of staff.

Everything remains the same. I wanted to just kind of cut over to you were in the main room this morning. Yes. Any impressions or any thoughts about anything you saw in there whether it was the IOT or the automation in Davis and

artificial intelligence? I've been looking at Davis since it came out as a beta version. Right. And it's something that I'm really interested in, and I'm trying to get a proof of concept for the state of Michigan. Okay.

The idea that you can have your monitoring applications pinpoint and resolve the issue for you. The self-healing concept, we just love that, and I'd like to get that in place. Right, and for me, seeing that recovery, I don't know if you were in there when Alois was presenting the voice operations. Were you in there towards the end of that bit when the app went down and then it came back up? Yes.

That I hadn't seen before. I didn't know we could do that. So it sounds like Davis is almost going to become a little bit of a staff multiplier for you. Right. It does the notifications for you, knows the appropriate people to contact with, depending on what type of alerts it receives. So it's just an awesome thing. Excellent. Anything else you, anything, any particular tracks that you're looking forward

to going and seeing today or anything you're looking to get out of your time here besides some nice weather, hopefully? Besides Davis, my primary attraction here is going to be the roadmap sessions, keeping up with where Dynatrace is going with their products, what they're doing as the manager for the team. I have to sell the stuff to the executives. Okay, excellent. I need to know what's going on. Excellent.

Thank you very, very much, Steve. We have a parting gift for you. You can have a PerfBite stopwatch. A PerfBite stopwatch. So you can time the performance the old-fashioned way. So if you get any functional tester that says, well, I'm not tracking time, hand them the stopwatch. Okay. Because if it doesn't work for one, it can never scale for many.

So if the Dynatrace server goes down, I have my backup. Exactly. My DR plan. Exactly. Thank you. Thank you very much. Thank you very much, Steve. Oh, and he's got a nice red hat hat.

It's a red hat. So Steve brought up an interesting point, the use of Davis as a staff multiplier effect. He did. Yeah. I don't know what to say. But, yes, absolutely. It can really do a lot of extra things for you. Particularly if you have a fixed team that it's difficult to add expertise or add personnel over time, using Davis to educate them about your infrastructure and then take advantage of automation, boom.

And even the alerting, too. I think that's one of the key things. Did you want to? Yeah. Automation, boom. And even the alerting, too. I think that's one of the key things. Did you want to? Yeah. And Davis can do it right now in real time when it happens instead of calling somebody and saying, find out where this guy's at and get this alert to him, which could take maybe 30 minutes to three hours later.

Fantastic. Excellent. So very proactive in this case. Yes. Thank you again. Thank you, sir. Have a good time. So I hadn't really thought about that.

I've been kind of myopic then, my perspective on AI, helping me, or AI helping make my job more efficient to go after patterns which are known and easy to address, have those either proactively addressed or notified when they come out, perhaps even as part of a DevOps where we can identify them in the unit testing and component assembly phases so they never escape the wild and get to production, I hadn't really considered the labor multiplier effect of I can't add staff or it's difficult for me to buy this expertise on the open market. Right, or not even you can't add staff.

Well, besides the staff, though, but think about if you're going into a microservices framework, right? You're going to do the whole cloud. Where it's auto-expanding, where it may start up 500 new nodes at a time. Where you take your monolithic app and break it down into 30, 40, 60 little services and apps, or however many, it could be hundreds, right? A company will never be able to hire the staff that can maintain that. And I don't mean developing-wise, but I mean monitoring, figuring out which pieces are going, going responding to all the different alerts because all that's just going to oh no i think operations would go for that budget but they may not get it but they're going to ask they're going

to ask for it right but it's going to be it's i think going forward it's going to be more and more of an exponential purchase so you know whether or not it's davis but hopefully it's davis yeah right you're going to need a digit there's going to have to be some machine helping out with that. Right? And I think that's one of the only ways that with the scale of some of these things going on, it's going to happen. You don't see Google having a staff of 20,000 people to monitor their operations team. Right? Because they have 500,000 servers or whatever and services running out there.

They have probably not a very large staff doing that because they have it set up where it's very efficient. That's right. They have all those lights out data centers. Right. That are solar powered and cooled by convection. Convection cooling. But it's more of as things become so complicated, are we going to start creating things that are beyond our brain capability to keep an eye on? Yeah, or not to envision but to even manage. Sometimes with new technology when it comes out, just alone it makes my head swim. What I don't want is I don't want Davis to become an architect for the next generation of application architecture,

because then Davis has guaranteed himself a job. That's it. Can't get rid of him now. He designed this thing. Only he knows how it works. Or what will be bad is if developers know, well, Davis can fix it in five minutes, so let's just push it. Oh, you know, I could see that happening. Oh, I can too. That goes back to the human nature of taking the path of least resistance to get it out the door.

And performance is hard. It really is. It's easy to build a functional interface for one person. But to build something that scales, scales elegantly, is resource efficient, takes quite a bit of element of design, which is going by the wayside in a lot of Agile and DevOps shops. It's get it out the door, throw it in production. If it fails, roll it back, and we'll have another build in three hours. Right. And there's really no excuse for that these days with all the tooling that you have to give you the feedback you need throughout that life through the pipeline. There's no reason to just put those checks and balances in place.

You're right. They absolutely should be at every stage of production putting it in place. Well, it is 10.48. Yeah, so is it about time for a break? Yeah, we can go ahead and take a break. Oh, let's actually stop. Let's grab the mic. Do you have two minutes, three minutes?

Get on the mic. So we're going to get an impression from... Yes, Florian. Florian. He is the... What is your... He's with Dinah Trace. What is your official title, Florian?

Oh, yeah. So... Over the years, I think I had many titles, but I'm with the team for 10 years now. So I'm one of the... I was actually the first product manager working for Dynatrace 10 years ago. And now what I'm called is the chief product officer.

I'm caring a lot about product, making customers successful, being easy, analytics, AI. So this is what I do. So we've been chatting about the kickoff conversation. None of which is a surprise to you, obviously. So you're looking at the next generation beyond what was discussed in the kickoff. Right. Is there anything that you can share with us, some exciting developments that are coming

that are publicly known but maybe not widely known? Absolutely. So maybe stick to something that is usually not getting the big stage at all. So talking about architecture, for example. So who cares about architecture, especially of an APM product? Many people do not care, but they care about the value add that can be getting out of this.

So, for example, we started to build a SaaS product, always keeping in mind that we also want to enable customers and organizations to install it on-premise and run their own SaaS business company internally. So offering APM to their internal customers as a service. So essentially building an application utility. Right. Where you need, of course, a lot of automation because if you're running a large company with 10,000 of engineers, then suddenly it becomes pretty boring to always ask for the IT team to get access to this app or to this product. So by using this SaaS approach also

for on-premise and enterprises, it's as easy as signing up for a SaaS trial online, but you do it locally in your on-premise secure data center. And this is really a game changer to big organizations because it makes their life so much easier. Right, right, right. So are you looking to deploy as like a VMware image or other deployment models that basically you can pick it up and drop internally, similar to how we can do it on Amazon and IBM?

It's about bringing APM to the masses. It is very important for any engineer, whenever there's a new piece of technology, that it's super easy to roll out. Ideally, it's not even a click that is required. Maybe one line of a command line on your Mac or on a Linux box,

that's actually enough. It has to be easy. This is what we are striving for. What have we been chatting about? Developers and humans being lazy? Yes. Path of least resistance? I remember when I was learning some early Unix,

which was based on Linux, and all these commands are really small and abbreviated because Unix developers didn't like typing a lot. That's just extended. Yes. Forever. Yeah. So we were actually just talking about a crazy thought for you.

Science fiction world, but maybe not. Developers liking to be lazy. Davis, what we just saw in that presentation was, what Alois just showed us was Davis fixed the problem, rolled back, and took care of it for you right the only fear i have with that is the developer saying well if there's a problem davis will roll it back in a minute so let's just push it to production did you have you did you did you have you ever had any thoughts of you might be enabling developers to to to forego performance again.

That's not the first time we're getting that question, actually. It's time to push Davis further around into the development cycle to catch those issues earlier so that they don't cascade forward into production. That's the right thought. So if you do agile and if you, whether it's a SaaS service or something that you simply make your customers install, doesn't really matter. If you've got to be agile, then you have this delivery pipeline. Right. And if there are hundreds of people, of engineers working on one service or product, then one of the key things you have to keep in mind is to have a pipeline that works, that scales up with the needs that your

engineers are having, and you need this quick feedback, right? So I guess the next natural question is Davis integration with Bamboo, Davis integration with Jenkins. To ask those complex questions, is it there today? It is there today. Oh, fantastic. Yes, it's on the robot.

So no excuses in a CI environment. And I can see it being like HAL 9000 where it can get, if the code went through all the checks and balances, there can be some machine-embedded line of code that can't be created by a human being somehow. Brian, what are you doing, Brian? It's bad code, Brian. When it tries to deploy it, I don't think you should deploy that. Hey, Brian, we are professionals, so there is nothing like bad code. We always get it right in the beginning.

Exactly, just like Alois did in his presentation. No bugs. The cool thing about working at Dynatrace really is that we are in the same boat, so we are not creating this solution where engineers are in a foreign domain. So think of creating an online banking service. Of course, you're a banking user, but the domain of finance is, of course, very different and is alienating what usually a software engineer is thinking about.

But what we do is in the heart of software engineering as well. So we can relate to those issues ourselves too. So getting a pipeline up and running really works well. It takes a lot of energy and time. So whatever tool support you need and you can get you actually get it. So this is the classical dog food engineering model.

Totally. No, you know what though? I don't know if it was Anita or Sonia. One of them had a much better thing to say instead of eat your own dog food. It was drink your own champagne. And I'm like, that is a much better way of putting it because why the heck would I want to eat dog food? I'll drink my own champagne.

Who wants to eat dog food? Right, exactly. Dogs. They probably don't even want to eat it, but that's what they get, right? The thing you said earlier about the Unix guys automating the crap out of everything and, well, are not into typing stuff anymore. I mean, I myself, I've been an engineer for

25 years now, and just remembering whatever system commands you need to do something. Yeah, and I need to pipe it from A to B to C and then I need to do the little carrot to output it to this location to be picked up. So you end up debugging the command line scripts.

And there's stuff going on that's crazy. Yeah, yeah. All right, well, thank you very much, Florian. Sure, guys. Thanks for having me. Maybe swing back tomorrow or something when anything else might be interesting that you'd seen or heard. Excellent. Sure, thanks, guys. Yeah, and if you'll stick around for just a moment, we've got a parting gift for you.

Oh. Yeah. I have all the time in the world. Oh, sure you do. Yes, you have nothing to do today. So let's close out this segment. Okay, excellent. We'll be back online in about 30 minutes with Dynatrace 2017 Perform.

With PerfBytes and Pure Performance. And Pure Performance. Thank you very much, everybody. Talk to you soon. We'll be back. All right. Thanks. Thank you.