Vignesh Shanmugam

Wrote a blog post on "Understanding Long Tasks data in Real User Monitoring" Hope it opens up new opportunities in web performance monitoring. @_vigneshh/understanding-long-tasks-data-in-the-real-user-monitoring-rum-world-6e74e5bdfcd6" target="_blank" rel="nofollow noopener">medium.com/@_vigneshh/und… #rum #webperf #javascript

@threepointone Lovely framing and great read, immediate thought comes to mind is what would happen if we feed this post to the coding agent 🤔

new post: where good ideas come from (for coding agents) sunilpai.dev/posts/seven-wa… thought about why some people cook with coding agents, and others bounce off. my take: llms are "thought completers" traversing idea-space - so good results come from good navigation: constraints, context, oracles, loops. I used steven johnson’s "where good ideas come" from as a rubric and mapped it onto coding agents. also a note about how "juniors" can adapt to this new feature, and what "creating software" looks like practically.

My kid played the "Jurassic Park theme song" youtu.be/PC2rc7qyJVQ #keyboard #music #jurrasicpark

🔥 The newest web compression API announced by @patmeenan at #PerfNow It could be up to 95% more efficient than existing gzip/brotli methods. All leading browsers agree on the spec => it will be shipped soon, and it's super exciting. Learn more: use-as-dictionary.com

This launch is super close to me personally as it allows the developers to measure the Web performance aspect while ensuring the website is functional. Ex: Super handy during functionality/load tests on peak season (black friday).

What I have been working on lately - Elastic Synthetics monitoring. You can write End-to-end synthetic suites in JS and run them on multiple pop's to understand the website performance. #synthetics #webperf #playwright elastic.co/blog/whats-new…

@PKodmad Congrats to you and your family, Truly amazing times.

I had our baby girl Kavana on Nov 20th. Feeling blessed and overwhelmed. Such a special bubble to be in amidst all the outside chaos.

Agree 💯, Copilot helps with rapid prototyping when your mind is empty and you are staring at your IDE. But damm the effort you put in to debugging because of the buggy code it produces 😅 twitter.com/sunshowers6/st…

Check out the blogpost we wrote on this architecture around 2016 oreilly.com/content/better…

Year 2015 coming back to my memories reading this blogpost - blog.cloudflare.com/better-micro-f… We were running similar architecture for almost 3-4 years at @ZalandoTech - mosaic9.org with lots of other goodies like caching, Observability, backoffs, pooling etc.

@_prateekbh Come to Seattle bro!

👋 Friends in SF. I am around for the weekend.. Hit me up if anyone's interested in getting a coffee together.

@siddharthkp @_prateekbh @ankeetmaini Only thing that helped while I worked at Zalando in the past 1. Tooling to figure out how the components are used by different teams - #slide=id.g1feee17fc3_1_34" target="_blank" rel="nofollow noopener">docs.google.com/presentation/d… 2. Planned updates - Each team has a dedicated owner who sort of helped to migrate.

@_prateekbh @ankeetmaini it’s been helpful, but not always possible example: api changes that change the composition or mixed with application behaviour

okay, here's some niche work stuff in a component library, should you version and publish each component as a separate package or publish the whole library together? sid.st/unpolished/com…

We’ve built some unusual new features for Elastic Synthetics 8.4.0, now we’re finally talking about them: youtu.be/lG4OcV314_k . Watch me and @justin_kambic discuss

@Zizzamia @coinbase Gotcha, makes total sense on the long tasks part. Does the 2second dynamically get adjusted if one of the JS resources takes a long time to load? Otherwise a page could cheat and delay the render and nothing gets measured. Do you tackle this case?

@_vigneshh @coinbase That's where we define the 2 seconds Navigation limit of Total Blocking Time as Good at 10% and Poor at 20%, respectively 200 and 400ms. That's on Web at least; for React Native, there is more Javascript involved, and the limit for Poor is at 30%, which is 600ms.

Some personal news ✨ The "Performance Vitals" article is out 🤩 A unified scoring system to guide performance health and prioritization at @coinbase. Oh, and I got promoted to Senior Staff Software Engineer 🥳🪩🍾 blog.coinbase.com/performance-vi…

@Zizzamia @coinbase Basically how you decided on the 2second mark for calculating the NTBT.

@_vigneshh @coinbase Thank you ✨ Are you referring to the difference between Web vs Mobile, or the mark that splits poor vs good?

@GergelyOrosz Ex Zalando engineer here. Zalando used to pay 150EUR/day back in 2017-2018 and it's per week (1050 EUR).

Going through to all the responses and DMs in "does your company pay for oncall? If so, how much?" Google pays the most of all I've seen so far, but I don't yet have the exact numbers (it's a complex formula, but generous payments). And the other highest ones I came across:

@_prateekbh @VivekNayyar09 Big in the UI itself on how it's calculating percentiles.

@_vigneshh @VivekNayyar09 Bug in what?

#datadog question: I have a metric being recorded for a month in datadog. How do i get P50 and P90 values for it for a week?

@_prateekbh @VivekNayyar09 Totally possible if your datapoints are way too low, but if that's not the case it could be a bug.

@VivekNayyar09 somehow median and 95percentile are giving me same result.. any ideas?

@delvedor @elastic Congrats Tomas 🎉

After almost four incredible years as IC, I'll transition to an engineering manager role from tomorrow! I'm so glad that I had/have the opportunity to work with such talented people at @elastic, and I can't wait to start this new adventure 🚀