text stringlengths 0 2.35k |
|---|
**Natalie Pistunovich:** But then I will include the question on the difference from mlops to aiops. \[laughter\] |
**Johnny Boursiquot:** Right... |
**Natalie Pistunovich:** Then I guess comes the fun part... Unless, Johnny, you want to ask anything before we start talking about the unpopular parts. |
**Johnny Boursiquot:** Let's get to the unpopular stuff. |
**Natalie Pistunovich:** Alright. |
**Jingle:** \[41:07\] to \[41:22\] |
**Natalie Pistunovich:** I don't know if I'm supposed to say this, and I'm sorry if not, but the performer of this wonderful song has a birthday today. Finally turning 18. So if you see Mat on Twitter... \[laughs\] Wish him a happy birthday. But only if you're listening to the live recording. Actually, you know what - ... |
**Johnny Boursiquot:** He's finally an adult. |
**Natalie Pistunovich:** \[laughs\] Maybe he'll grow up. |
**Johnny Boursiquot:** Maybe he'll grow up. \[laughs\] That's an unpopular opinion, I guess. Mat, we love you very much. It's all in good vibes. Mike, do you have an unpopular opinion for us? |
**Mike Eastham:** Yes, I have an unpopular opinion. It's not tech-related, which I hope is okay, but... |
**Johnny Boursiquot:** It's all good. |
**Natalie Pistunovich:** That's not popular. |
**Mike Eastham:** I really just don't like maple syrup, is my unpopular opinion. Even on pancakes, on waffles - I just don't do it. That's my opinion. |
**Johnny Boursiquot:** Okay... |
**Natalie Pistunovich:** Johnny, I saw a tweet from you from the weekend that you were making pancakes. |
**Johnny Boursiquot:** Yeah, yeah. You know, I have young kids, and they love themselves a pancake, especially the syrup, so... \[laughs\] |
**Mike Eastham:** It seems almost universal... Just not for me. |
**Natalie Pistunovich:** What don't you like about it, Mike? |
**Mike Eastham:** How sticky it is is a big part of it. |
**Natalie Pistunovich:** Is it more or less sticky than honey? |
**Mike Eastham:** Somehow it feels more to me... I don't know if that's chemically accurate, but that's my perception anyway. And we've got a lot of Canadians working at Tecton actually; in fact, the CEO is Canadian, so... I hope if they hear this I don't get ostracized. |
**Johnny Boursiquot:** \[laughs\] They'll be like, "Oh, you don't like syrup, huh?" Well, there you have it. Unpopular opinion. |
**Natalie Pistunovich:** Yeah. I think the most funny opinions we have is in the context of food, so... Great choice for an unpopular opinion. There will be a Twitter survey; we will see how unpopular is your unpopular opinion. You might get into the top five, or low five |
**Mike Eastham:** Are you supposed to have the most people disagree with you? |
**Johnny Boursiquot:** I mean, it's a goal... We all have life goals, so why not this one? |
**Mike Eastham:** Okay. Hopefully I do well then. |
**Johnny Boursiquot:** Mm-hm. |
**Natalie Pistunovich:** Cool. Alright, well that was a very interesting conversation, and I think it's a rather new field, at least in our context of gophers, this mlops in general, but also feature stores in specific... And we will be providing links for you to learn if you wanna dive into that, or if you wanna contr... |
**Johnny Boursiquot:** Indeed. |
**Mike Eastham:** Thank you for having me. |
**Natalie Pistunovich:** Thanks, everyone. |
**Johnny Boursiquot:** Alright, y'all. |
• Making code efficient in Go |
• Premature optimization vs. optimization later |
• Balance between readability/maintainability and efficiency |
• Importance of understanding problem space and team abilities before optimizing |
• Use of tools for measuring and benchmarking code efficiency |
• Donald Knuth's quote on 97% of the time not requiring optimization |
• Challenges in identifying opportunities for efficiency improvements |
• Importance of measurement and observation to identify areas for improvement |
• Role of observability in identifying bottlenecks and performance issues |
• Measuring and monitoring resource usage and time it takes for tasks |
• Identifying key metrics (e.g. HTTP request latency) and drilling down into contributing factors |
• Using the same metric for guidance on code improvements as SREs/operators use |
• Importance of continuous profiling and monitoring to optimize system performance |
• Integrating observability signals with instrumentation to improve efficiency |
• Latency metrics and SLO-based alerts for optimizing performance |
• Production assessment and benchmarking as part of the development process |
• Comparing micro and macro benchmarks in different environments (CI, local machine, production) |
• Benefits and challenges of collecting metrics in production versus development environments |
• Techniques for managing tech debt and prioritizing optimization efforts |
• Balancing features and performance in software development |
• Managing tech debt and optimizing resource usage |
• Setting goals and budgets for performance requirements |
• Understanding the cost of observability and logging |
• Trade-offs between features and performance in decision-making |
• Using metrics, tracing data, and profiling to measure system efficiency |
• Discussion on the importance of considering cost when implementing features, such as deletion requests |
• Using time.sleep intentionally to slow down services to avoid overloading or to give users realistic expectations |
• The potential for controversy if users find out that a service is deliberately slowed down |
• Observability and profiling in Go, including metrics and pprof (Google's profiling tool) |
• Go's built-in support for observability and profiling as a benefit compared to other languages |
• Go's standard logging library is being improved but has controversy around it |
• Observability and introspectability are key considerations for the Go community |
• Interrupt-based profilers have a cost and can be intrusive |
• eBPF (Extended Berkeley Packet Filter) offers a new solution for observability tooling with minimal cost |
• eBPF allows for profiling without instrumenting code, with potential applications in security and network tooling |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.