text stringlengths 0 2.02k |
|---|
• They aim to have a consistent and regular format for sharing stories and experiences, and to invite other teams and experts to share their own stories |
• The hosts acknowledge the complexity and ever-changing nature of infrastructure and technology, and the need for ongoing learning and exploration |
• Elon Musk as a guest on Ship It to discuss shipping Kubernetes to Mars |
• Ship It's approach to improving infrastructure and speed |
• The importance of knowledge and understanding in improving systems |
• Potential community involvement in the show, with topics and guests suggested by listeners |
• Gerhard's goal of sharing interesting topics and solving specific problems that others would find helpful and interesting |
• The possibility of a Ship It community, with enthusiasts sharing ideas and approaches on how to improve systems |
• Creation of a community channel (Ship It) for discussion and idea sharing |
• Comments enabled on episodes, with potential for improvement in design and user experience |
• Discussion of community engagement and feedback mechanisms (e.g. Changelog.com/request) |
• Importance of building systems for easy and straightforward shipping |
• Introduction of Gerhard Lazu as a weekly podcast host for Ship It |
**Jerod Santo:** We're back with Gerhard Lazu, our resident SRE. What's up, Gerhard? |
**Gerhard Lazu:** It's all good. It's actually ten times better... |
**Jerod Santo:** Our website is? I hope so... |
**Gerhard Lazu:** That's the title of this show, it's "Ten times better." \[laughter\] |
**Adam Stacoviak:** I like ten times anything. |
**Jerod Santo:** Yeah, 10x SRE, or what's going on here? |
**Gerhard Lazu:** That's exactly what it is. It's 10x... That was the theme for this setup; it has to be ten times something. It doesn't matter what that ten times is, as long as it is ten times something, like an order of magnitude better. And guess what - it is. |
**Jerod Santo:** Nice. So it could have been ten times slower to deploy, or ten times longer response times - none of that. It had to be ten times better. |
**Gerhard Lazu:** Yeah. |
**Jerod Santo:** Well, for those who haven't listened to the annual Changelog infrastructure episode - welcome. You are here, this is not -- it hasn't been a whole year, it's been a half year, so it's now like a semi-annual... But we worked faster this time, didn't we, Gerhard? |
**Gerhard Lazu:** We did, because we had the basics covered really well, and the base was so good that iterating was super-simple, and what we iterated on was basically what mattered the most - uptime and response latency. We had a couple of tricks up our sleeve; I think it was combined, I had one and you had one... We... |
**Adam Stacoviak:** Fine-tuning. There's details in the fine-tuning that make things faster, and that's what you've gotta optimize for. |
**Gerhard Lazu:** Yeah, I think it takes a while to learn your system, I suppose, to learn all the components, properly learn them... And then when you're comfortable with all the components, figure out which is the smallest change that you can make for the biggest improvement. And that's what we did. |
**Jerod Santo:** Yeah. |
**Gerhard Lazu:** Shall we spoil it? I mean, if someone just wants to listen to five minutes, we can spoil it and they can-- |
**Jerod Santo:** No, let's tease it. Let's hold it back. |
**Gerhard Lazu:** Alright, we'll tease it. |
**Jerod Santo:** Stick around, listener... |
**Adam Stacoviak:** Yeah. |
**Jerod Santo:** \[03:49\] Let's start with this - not much changing this time around. A lot changed last time around. Our 2020 episode, which came out last October, was a big change... A lot going on, and some of the reaction to that episode was -- and we're on Kubernetes now, and it's like "Hey guys, you're on a thre... |
So let's start there, Gerhard... What do you think about that? Do you agree with that? |
**Gerhard Lazu:** Not really. \[laughter\] And this is like the \[unintelligible 00:04:22.23\] controversial part. |
**Jerod Santo:** I assumed you were gonna say that, because you're the one who set it up, so... |
**Gerhard Lazu:** Right, so... I think that's a very simplistic view, because you're right - when you boil it down, that's exactly what you have. It's just a Phoenix app, it's a web app. You have a database, you have a proxy (maybe), and that's about it. That's what you have. But it's almost like the iceberg. There's l... |
So what else do we have below? Well, you have certificates. You have load balancers. You have DNS. You have code updates. You have tests. You have CI/CD. You have dependencies. You have dependencies of dependencies. And the list goes on and on and on. And things are changing all the time. So given you have so many thin... |
What about the monitoring? How do you manage the monitoring? And again, it just goes from there, because you're running a production system. A production system that is serving a lot of traffic, which Changelog.com does. And even if it's a simple app -- I think it's simple because we made deliberate choices; it could b... |
**Jerod Santo:** Do you already know all the Kubernetes stuff? So when people think about setting up a Kubernetes cluster, they talk about the complexities of the API perhaps, or the tooling, or the ecosystem... I always think back to CNCF's -- it's not a roadmap; what is that? It's like a trail... |
**Gerhard Lazu:** The landscape. |
**Jerod Santo:** Yeah, the landscape... And there's all of these words that I don't know any of them, and each one of those is like a complex piece of software. And I get overwhelmed. You've got this rolled out, and I'm just curious, was there a Kubernetes learning curve for you, or had you already done that previously... |
**Gerhard Lazu:** So I had some knowledge, but it was mostly basic. But the thing to understand is that I have been doing infrastructure for - I don't wanna say decades, because that's like bragging... But let's put it "a really long time." We were joking about webmasters... I used to be one. CGI-bins - oh yes, baby. T... |
**Jerod Santo:** "Oh yes, baby..." \[laughs\] I remember CGI-bins. I wouldn't describe them as the good old times, but... You were probably better at them than I was. |
**Gerhard Lazu:** \[08:14\] Well, there you go... Perspectives. \[unintelligible 00:08:15.00\] and you remember the past much better than it actually was. So there's an element of that. |
**Jerod Santo:** Right. |
**Gerhard Lazu:** So I've been doing this for a really long time, and I can appreciate the cycles that we went through. We had many, many cycles, and I've learned to learn on the job. And if you optimize for that, there's nothing new that is too daunting, or too -- I mean, it's exciting; you'll make mistakes, but after... |
**Jerod Santo:** That sounds better. |
**Gerhard Lazu:** Uh, what is it...? |
**Adam Stacoviak:** Law of diffusion of innovation. |
**Gerhard Lazu:** That's it. |
**Jerod Santo:** What's that? |
**Gerhard Lazu:** So early adopters -- basically, any new thing, whenever you're introducing it, you have to focus on the first 2.5%, the early adopters. |
**Jerod Santo:** Oh, this is like the curve of people who are gonna adopt it. It starts with like the enthusiasts, and it goes to the... |
**Gerhard Lazu:** Exactly. Early majority... |
**Adam Stacoviak:** The spread of a new idea. |
**Gerhard Lazu:** Exactly. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.