text stringlengths 0 2.35k |
|---|
**Mike Fridman:** So I would have thought everybody would have agreed that we needed a logging interface, but it seems to be slightly more unpopular than I would have expected, so I'm actually genuinely curious what the feedback and the reaction would be... Because I've struggled with that quite a bit. |
**Jon Calhoun:** I'm assuming that you're in a place -- like, when you talk about a company that's gonna be using it in 15 projects, to me that sounds like a reasonably-sized company, with a decent amount of engineers... Whereas I think part of the reason that -- it's not that I disagree with that, it's just more it's ... |
**Mike Fridman:** Yeah... But it also creates a little bit of fragmentation, because if you have everyone redefining what an interface is within their project, I feel like that's something that should inevitably be solved at the standard library level... But who knows. |
**Jon Calhoun:** \[44:18\] It's a tough one. I guess we'll see. I suppose it's something that could potentially introduce a new package in the future for, and at that point third-party libraries would hopefully start to move towards that... But who knows. |
So I have a follow-up, maybe unpopular type of question for you... What do you guys think of the any alias that they're adding in Go 1.18? For anybody listening that doesn't know what this is, basically they're adding a type alias where the word "any" is going to map to the empty interface. So interface, then left curl... |
**Vojtech Vitek:** I don't have a strong opinion on this one. I'm quite used to any from TypeScript, even though if you're using TypeScript, you shouldn't be using any... Right? So I think it's fine; but yes, I get your point, because suddenly we will see interfaces and anys everywhere, and it will get mixed up. |
**Mike Fridman:** Yeah... I also don't have much of a strong opinion on that one. I installed 1.18, saw a whole bunch of errors with empty interfaces and any, I'm like "I'm going back to 1.17.5 and call it a day." |
**Jon Calhoun:** I used it briefly, and I will say that getting used to typing "any" was weird. I'd have to use it more to really get it... And I do agree that -- hopefully, a lot of projects just switch everything over to any, because without doing that... You're still gonna have that confusion of somebody new to Go e... |
**Mike Fridman:** That's such a good point, because one of the beautiful things about Go is I could go type something in a search engine, find an article from 2015 which would be 6-7 years old now, and it would be exactly the same. Not much has changed in Go. So there's so much resources out there that are gonna refere... |
**Vojtech Vitek:** I wonder if the go fmt will eventually replaces interfaces with anys. |
**Jon Calhoun:** It would be a nice way to clean it up, but I don't know... I guess we'll have to wait and see. I know I've seen people show some -- I think it was Brad Fitzpatrick... I think that's who it was; I'm trying to remember for sure... But I think he had posted something on Twitter that was like a short littl... |
So things like that... But still, like Mike said, with tutorials it's not gonna be that easy, which is gonna be kind of challenging. And that's gonna make it different, I suppose. I know that was one of the things that always frustrated me with JavaScript, is it felt like every few years there was kind of a new way to ... |
So I'm curious to see how it goes... I'm hopeful, but at the same time I suspect it might also lead to more confusion, for at least like the first year, where people are coming and they're like "Why are both of these things doing the same thing?" |
Okay, thank you, Mike, thank you, Vojtech for joining me. I will play us out. |
**Vojtech Vitek:** Thanks for inviting me. |
• Introduction and welcome |
• Site reliability engineering (SRE) and microservices |
• Nayana Shetty's experience with SRE at the Financial Times and Lego Group |
• Challenges in managing multiple monitoring tools and platforms |
• Importance of thinking about future sustainability and observability when building products or systems |
• The importance of designing systems with observability in mind from the beginning |
• The "carrot and stick" approach, where motivation (carrot) is used to encourage good design, while necessary measures (stick) are taken if not implemented correctly |
• The challenges of retro-fitting observability into existing systems |
• Designing for failure as a key aspect of building resilient systems |
• The value of testing and error handling in code, and the importance of thinking about potential failures |
• The need to balance observability with system complexity and criticality |
• The role of teams, technology, and culture in adopting observability best practices |
• Establishing criticality of an app and deciding on monitoring levels based on business needs |
• Using methods like RED (rate, error, duration) and USE (utilization, saturation, errors) for monitoring |
• Considering cost as the primary factor in monitoring investment decisions |
• Importance of central teams pushing value-add features to individual teams |
• Communicating and driving adoption of centralized monitoring through benefits and ease-of-use |
• Intrinsic value beyond achievement should be shown to teams |
• Carrots and ease of use are necessary for adoption |
• Central teams should promote tools and benefits to product teams |
• Education on tool value is crucial |
• Continuous evolution and monitoring is necessary |
• Focusing solely on delivering tools can lead to lack of adoption |
• "You build it, you run it" approach helps with understanding customer needs |
• Code should be debuggable at any time |
• The challenges of practicing skills in a drill-like environment |
• Value of drills and shadowing in team settings for problem-solving and learning |
• Radical Candor workshops as an example of practicing feedback and communication skills |
• Alert fatigue: the phenomenon of ignoring too many alerts due to noise and overwhelm |
• Strategies for managing alert fatigue, including auditing and cleaning up unnecessary alerts |
• Ideas for incentivizing teams to clean up and manage their alerts effectively |
• Discussion around collecting emojis and integrating them into a Grafana dashboard |
• Transferable skills in monitoring and observability across different organizations |
• Benefits of using open-source tools for monitoring, including ease of adoption and community support |
• Importance of considering the organization's landscape and introducing new concepts incrementally |
• Nayana Shetty's approach to monitoring: focusing on core aspects (logging, metrics, alerting, tracing) and evaluating capabilities rather than specific tools |
• Evaluating market trends and innovation in the monitoring space, with mentions of Loki, Prometheus, Graphite, and Grafana |
• Prioritizing problem-solving over tool-specific solutions |
• Nayana Shetty's speaking style and experience |
• Sending messages to past selves vs future selves |
• Matt Toback and Mat Ryer's conversation about the space-time continuum |
• Upcoming conferences and events for Nayana Shetty |
• Grafana's Big Tent podcast format and topics |
• Personal conversations between hosts and guests, including a discussion about music and playing an instrument |
**Mat Ryer:** Hello, and welcome to Grafana's Big Tent, the podcast all about the people, community, tools and tech around observability. I'm joined today by - it's only Matt Toback. Hello, Matt. |
**Matt Toback:** Hi, Matt. What do you mean "only"? |
**Mat Ryer:** Yeah, no, it's just understated; it's just like "I can't believe it's you, really, in a lot of ways..." |
**Matt Toback:** It's how you say it, it's not what you say. "I can't believe it's you! You're here!" |
**Mat Ryer:** Oh, right. Okay. I can't believe it's you...! |
**Matt Toback:** You're like, "You're here... You're here." |
**Mat Ryer:** Yeah. Well, don't worry, it's not just me and you. That would be obviously tiresome for both of us. We have a special guest joining us today. |
**Matt Toback:** That would be tiresome for both of us! |
**Mat Ryer:** That's how you say it. You're right. \[laughter\] Today we're joined by Nayana Shetty from Lego. Hello, Nayana. |
**Nayana Shetty:** Hi, Mat, and Matt. \[laughter\] |
**Mat Ryer:** Yeah, just one would suffice. We'll share it. Happy to share. And you are the principal engineer who loves talking about SRE and microservices, right? |
**Nayana Shetty:** Yes. I think over the years I've been in teams where we've built microservices, and when you scale up and have hundreds of microservices, how do you then make them reliable and keep them reliable? That's what I'm interested in. |
I was working at the Financial Times, where we had all of these hundreds of microservices, and how do we manage it there. And now I've moved to the Lego Group, where we're going through massive digital transformation. And here it's like we want to build these hundreds of microservices, so should we care about reliabili... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.