 Good afternoon, cloud community, and welcome back to beautiful Las Vegas, Nevada. We are here at AWS to reinvent day four of our wall-to-wall coverage. It is day four in the afternoon, and we are holding strong. I am Savannah Peterson, joined by my fabulous co-host, Paul Gillan. Paul, how you doing? I'm doing fine, Savannah. You look great. We're in the home stretch here. Yeah, we are in the home stretch. You still look fresh as a daisy. I don't know how you do it. You're too kind, you're too kind. I'm vain enough to take that compliment. I'm very excited about the conversation that we're going to have up next. We get to get a little DevRel, and we got a little swagger on the stage. Welcome, Austin, how you doing? Hey, great to be here. Thanks for having me. Yeah, it's our pleasure. How's the show been for you so far? Busy, exciting, feels a lot like it used to be, right? Yeah, I know. A little reminiscent of the before times. Before times. Before we dig into the technical stuff, you're the most intriguingly dressed person we've had on the show this week. I feel extremely underdressed. Well, so, and we were talking about developer fancy. Talked to me a little bit about your approach to fashion. Wasn't expecting to lead with this, but I like this actually. Thanks for going to my PR blog, I love it. My approach, here's the thing. I give free advice all the time about developer relations, about things that have worked and don't work, and community, and all that stuff, and I love talking about that. Someone came up to me and said, where do you get your fashion tips from? What's the secret Discord server that I need to go on? I'm like, I will never tell. Oh, okay. This is an actual trade secret. Top secret, wow, talk about. If someone else starts wearing the hat, then everyone's going to be like, there's so many white guys in, look, I'm a white guy with a beard that works in technology. I've never met one of those. Exactly, there's none of them at all. So, you have to do something to kind of stand out from the crowd a little bit. I love it, and it's a talk trigger. We're talking about it now. We're talking about it now. Oh, yeah. The production team loved it. It's fantastic. It's great, it's great. So, your DevRel for LightStep. In case the audience isn't familiar, tell us about LightStep. So, LightStep is a cloud-native observability platform built at planet scale, and it powers observability in some places you've heard of, like Spotify, GitHub, right? And we're designed to really help developers that are working in the cloud with Kubernetes, with these huge distributed systems, understand application performance and being able to find problems, fix problems. We're also part of the ServiceNow family, and as we all know, ServiceNow is on a mission to help the world of work work better by powering digital transformation around IT and customer experiences for their many, many, many global 2000 customers, and we love them very much. It's a big love fest here. A lot of people have talked about the collaboration. So many companies working together. You mentioned unified observability. What is unified observability? So, if you think about a tradition, or if you've heard about this traditional idea of observability where you have three pillars, right? You have metrics, and you have logs, and you have traces, and all those three things are different data sources. They're picked up by different tools. They're analyzed by different people for different purposes. And what we believe and what we're working to accomplish right now is to take all that, and if you think about these pillars, flip them on their side and think of them as streams of data. And if we can take those streams and integrate them together and let you treat traces, and metrics, and logs, not as these kind of inviolate experiences where you're kind of paging between things and going between tab A to tab B to tab C and give you a standard way to query this, a standard way to display this, and letting you kind of find the most relevant data, then it really unlocks a lot of power for developers and SREs to spend less time like managing tools, figuring out where to build their query or what dashboard to check, and more just being able to kind of ask a question, get an answer. And when you have an incident or an outage, that's the most important thing, right? How quickly can you get those answers that you need so that you can restore system health? And you don't want to be looking in multiple spots to figure out what's going on. Absolutely. Some people here are unified with their ability and they go to tool consolidation, right? And that's something I hear from a lot of our users and a lot of people just at re-invent. I'll talk to SREs, they're like, yeah, we've got six or seven different metrics products alone, just on services that they cover, and it is important to kind of consolidate that, but we're really taking it a step lower and we're looking at the data layer and trying to say, okay, if the data is all consistent and vendor-neutral, then that gives you flexibility not only from a tool consolidation perspective, but also a consistency, reliability. You could have a single way to deploy your observability out, regardless of what cloud you're on, regardless if you're using Kubernetes or Fargate or whatever else, or even just bare metal or EC2 bare metal, right? There's been so much historically in this space, there's been a lot of silos. And we think that unifying the observability means that we kind of break down those silos, right? The way that we're doing it primarily is through a project called OpenTelemetry, which you might have heard of. You want to talk about that a minute? Yeah, let's talk about it right now. Why don't you tell us about it? Keep going, you're great. You're on a roll. I am, just vamping. We'll just hang out over here. It's day four. I'm going to ask the questions and answer the questions. Yeah, Steve Gray. I do do, yeah. OpenTelemetry. OpenTelemetry. Explain what OpenTelemetry is first. OpenTelemetry is a CNCF project, a cloud-native computing foundation, and the goal is to make high-quality telemetry data a built-in feature of cloud-native software. So right now, if you wanted to get logging data out, depending on your application stack, depending on your application run time, depending on language, depending on your deployment environment even, you might have to make a lot of choices about like, what am I going to use? So many different choices, and the players are changing all the time. Exactly. And a lot of times what people will do is they'll go and they'll say like, well, we have to use this commercial solution because they have a proprietary agent that can do a lot of this for us. And if you look at all those proprietary agents, what you find very quickly is it's very commodified. There's no real difference in what they're doing at a code level. And what's stopped the industry from really adopting a standard way to create this logs and metrics and traces is simply just the fact that there was no standard. And so OpenTelemetry is that standard, right? We've got dozens of companies, many of them like very, many of them here, right? Competitors all the same working together to build this open standard and implementation of telemetry data for cloud native software and really any software, right? Like we support over 12 languages. We support, you know, Kubernetes, Amazon, AWS is a huge contributor actually, and we're doing some really exciting stuff with them on their Amazon distribution of OpenTelemetry. So it's been extremely interesting to see it over the past like a couple of years ago from like, hey, here's this like new thing that we're doing over here to really is a generalized acceptance that yeah, this is the way of the future. This is what we should have been doing all along. Yeah. My opinion is there is a perception out there that observability is kind of a commodity now, that all of the players have the same set of tools, same set of 15 or 17 or whatever tools, and that there's very little distinction between in functionality. Would you agree with that? I don't know if I would characterize it that way entirely. I do think that there's a lot of duplicated effort that happens and part of the reason is because of this, you know, telemetry data problem, right? Because you have to wind up, you know, there's this idea of table stakes monitoring that we talk about, right? Table stakes monitoring is the stuff that you're having to do every single day to kind of make sure your system is healthy, you know, to be able to, when there's an alert that gets triggered, to see why it got triggered and to go fix it, right? And because everyone has to kind of work on that table stake stuff and then build all these integrations, there's very little time for innovation on top of that, right? Because you're spending all your time just like working on keeping up with technology. Doing the boring stuff to make sure that we also follow up. Right. Yeah. What I think the real evangelic telemetry is is that it really, from like a vendor perspective, like it unblocks us from having to kind of do all this repetitive, commodified work that lets us help move that out to the community level so that, you know, instead of having to kind of build your Kubernetes integration, for example, you can just have like, hey, open telemetry is integrated into Kubernetes and you just have this data now. And if you are a commercial product or if you're even someone that's, you know, interested in fixing, scratching a particular itch, right, about observability. And it's like, I want to, I have this specific way that I'm doing Kubernetes and I need something to help me really analyze that data. Well, I've got the data now. I can just go create a project. You know, I can create an analysis tool. And I think that's what you'll see over time as open telemetry promulgates out into the ecosystem is more competent, you know, more people building interesting analysis features. People, you know, using things like machine learning to analyze this large amount of open, you know, large and consistent amount of open telemetry data. If it's going to be a big shakeup, I think, but it has the potential to really unlock a lot of value for our customers. Well, so you're a developer relations guy. What are developers asking for right now out of their, out of their observability platform? It's a great question. I think there's two things. The first is that they want it to just work. It's actually the biggest thing, right? There's so many kind of, this goes back to the tool proliferation, right? People have too much data in too many different places and getting that data out can still be really challenging. And so the biggest thing they want is just like, I want something that I can, I want a lot of these questions I have to ask answered already and open telemetry is going towards it. You know, keep in mind, the project's only three years old. So we obviously have room to grow, but there are people running it in production and it works, you know, really well for them. But there's more that we can do. The second thing is, and this isn't what really is interesting to me is it's less what they're asking for or more what they're not asking for because a lot of the stuff that you see people, you know, saying around, oh, we need this like very specific sort of lower level telemetry data or we need this kind of universal thing. People really just want to be able to get questions or get questions answered, right? They want tools that kind of have these workflows where you don't have to be an expert because a lot of times this tooling gets locked behind sort of, it's gate kept almost in a organization where there are teams that's like, we're responsible for this and we're going to set it up and manage it for you and we won't let you do things outside of it because that would mess up, you know. Here's your sandbox. Right, this is your sandbox you can play in and a lot of times that's really useful and very tuned for, you know, the problems that you saw yesterday, but people are looking at like what are the problems I'm going to get tomorrow, right? We're deploying more rapidly. We have more and more intentional change happening in the system. Like it's not enough to have this reactive sort of approach where our SRE teams are kind of like, or this observability team is building a platform for us. Developers want to be able to get in and have these kind of guided workflows really that say like, hey, here's where you're starting at. Let's get you to an answer. Let's help you find the needle in the haystack as it were without you having to become a master of six different or seven different tools. Right, and it shouldn't be that complicated. It shouldn't be, I mean, we've certainly, you know, we've been working on this problem for many years now, starting with a lot of our team that started at Google and helped build Google's, you know, planet scale monitoring systems. So we have a lot of experience in the field and it's actually an interesting story that our founder or now general manager tells BHS, Ben Siegelman, and he told me this story once and it's like he had built this really cool thing called Dapper that was a tracing system at Google. And people weren't using it because they're like, this is really cool but I don't know how to, but it's not relevant to me, right? And he's like, the one thing that we did to get to increase usage 20 times over was we just put a link. So we went to the place that people were already looking for that data and we added a link that says, hey, go over here and look at this, right? And it's those simple connections being able to kind of draw people from like point A to point B, take them from familiar workflows into unfamiliar ones, you know, that's how we think about these problems, right? How is this becoming a daily part of someone's usage? How is this helping them solve problems faster and really improve their life? Yeah, exactly, it comes down to quality of life. Wernher Frugl's made the case this morning that computer architecture should be inherently event driven and that we are moving toward a world where the person matters less than what the software does, right? The software is triggering events. Does this complicate observability or simplify it? I think that at the end of the day it's about getting the observability to me in a lot of ways is about modeling your system, right? It's about you as a developer being able to say, this is what I expect the system to do. And I don't think the actual application architecture really matters that much, right? Because it's about you, you are building a system, right? It can be event driven, it can be, you know, simple request responsive, whatever it is, you have to be able to say, this is what I expect to have. For these given inputs, this is the expected output. Now maybe there's a lot of stuff that happens in the middle that you don't really care about. And then I talk to people here and everyone's talking about serverless, right? You can see there's obviously some amazing statistics about how many people are using Lambda. And it's very exciting. And yeah, there's a lot of stuff that you shouldn't have to care about as a developer. But you should care about those inputs and outputs and you will need to have that kind of intermediate information to understand what was the exact path that I took through this evented system? What was the actual resources that were being used? Because even if you trust that all this magic behind the scenes is just going to work forever, sometimes it's still really useful. I have that sort of lower level abstraction to say like, well, this is what actually happened so that I can figure out when I deployed a new change, did I make performance better or worse? Or being able to kind of segregate your data out and say like, doing A-B testing, right? Doing canary releases. Doing all of these things that you hear about as best practices or well architected applications. Observability is at the core of all that. You need observability to kind of ask any of those higher level interesting questions. We are here at Reinvent. Tell us a little bit more about the partnership with AWS. So I would have to actually probably refer you to someone at service now on that. I know that we are a partner, we collaborate with them on various things, but really at LightStep, we're very focused on kind of the open source part of this. So we work with AWS through the open telemetry project on things like the AWS distribution for open telemetry, which is really it's open telemetry again, is really designed to be like a neutral standard. But we know that there are going to be integrators and implementers that need to package up and bundle it in a certain way, right? To make it easy for their end users to consume it. So that's what Amazon has done with ADOT, which is the short name for it. So it's available in several ways. You can use it as like an SDK and drop it into your application. There's Lambda layers. If you want to get Lambda observability, you just add this extension in and then suddenly you're getting open telemetry data on the other side. So it's really cool. It's been a really exciting to kind of work with people on the AWS side over the past several years. And I've personally seen just a lot of change. And I was talking to a PM earlier this week. I was like, hey, two years ago, I came and talked to you about open telemetry and here we are today, still talking about open telemetry. And they're like, what changes our customers have started coming to us asking for open telemetry. And we see the same thing now. Timing is right. Timing is right. But we see the same thing, even talking to ServiceNow customers who are these very big enterprises, banks, finance, healthcare, whatever, telcos. It used to be you'd have to go to them and say, let me tell you about distributed tracing. Let me tell you about open telemetry. Let me tell you about observability. Now they're coming in and saying, yeah, so we're standard, if you think about Kubernetes and how a lot of enterprises have spent the past five, six years standardizing on Kubernetes as a way to deploy applications or manage containerized applications, they're doing the same journey now with open telemetry. We're then saying, this is what we're betting on and we want partners, we want people to help us go along that way. I love it. And they're working in all CNCF projects as well that you're talking about. Right, so we're, yeah, we're integrated into Kubernetes where you can find open telemetry and things like Keptin, which is application standards and over time it'll just promulgate out from there. So it's really exciting time. A bunch of CNCF projects in this area, right? Prometheus. Prometheus, yeah. Yeah, so we interoperate with Prometheus as well. So if you have Prometheus metrics then open telemetry can read those. It's a, open telemetry metrics are like a super set of Prometheus. We've been working with the Prometheus community for quite a while to make sure that there's really good compatibility because so many people use Prometheus, you know? All right, so last question. New tradition for us here on theCUBE. We're looking for your 30 second hot take Instagram reel. Biggest theme, biggest buzz for those not here on the show floor. Oh gosh. And it could be for you too. It could be whatever. Yeah. I think the two things that are really striking to me is one, serverless, like I've seen, I thought people were talking about service a lot and they're talking about it more than ever. And two, I really think it is observability, right? Like we've gone from observability being kind of a niche thing. Not that you're biased. Yeah, huh? Not that you're biased. I'm biased, you know? It used to be a niche, I'd have to go, you know, a niche thing where I would go and explain what this is to people and now, you know, people are coming up, it's like, yeah, yeah, we're using open telemetry, using open telemetry. And it's very cool. It's a product, you know, I've been involved with open telemetry since the jump, since 20, since it was started really. And it's been very exciting to see and gratifying to see like how much adoption we've gotten even in a short amount of time. Yeah, absolutely. It's a pretty, yeah, been a lot. That was great. Perfect sound buy for us. Thanks, I love sound bites. Yeah. Awesome, we love your hat and your sound bites equally. Thank you so much for being on the show with us today. Thank you for having me. Hey, anytime, anytime. Will we see you in Amsterdam? Speaking of the KubeCon? Yes! Awesome, we'll be there. And there's some real exciting open telemetry stuff coming up for KubeCon, so. Ooh, well, we'll have to get you back on the KubeCon. Let it on the schedule. I love that for us. Thank you all for tuning in to our Wild'A'Wall coverage here, day four at AWS re-invent in fabulous Las Vegas, Nevada. With Paul Gillan, I'm Savannah Peterson and you're watching the Kube, the leader in high tech coverage.