nate berkopec / is on Nostr

Author, The Complete Guide to Rails Performance. Co-maintainer of Puma. http://speedshop.co (he/him)

Last Notes

2025-04-07 17:04:01

I'm holding a sale for April only on The Speedshop retainer: 20% off for your first 6 months. We've had ~50% of the first batch of companies renew so we've only got 3 slots open for the summer, claim yours now! Let's make your Ruby app fast!
https://www.speedshop.co/retainer.html

2025-04-02 17:04:01

It's often possible to "convert" an engineering problem into a money problem: "we can solve this by throwing more money/resources at it."
This is actually incredibly useful! Businesses love solving problems this way because these are the easiest problems to solve!

2025-02-13 02:17:44

Great work from Aaron on speeding up FFI (and potentially other things) in Ruby.
https://railsatscale.com/2025-02-12-tiny-jits-for-a-faster-ffi/
I want to call out the bit about Chris Seaton putting this idea in his head. Chris put a lot of great ideas in people’s heads, it’s one of the things he was best at. I miss him.
https://files.mastodon.social/media_attachments/files/113/994/174/501/297/764/original/d4f5d69ce125366d.jpeg

2024-12-26 22:04:57

Noah Gibbs has passed away.
Noah was:
* A member of the YJIT team.
* The author of "Rebuilding Rails"
* The creator of the Rails Ruby Bench.
Noah's passing was sudden and very fast. He did not suffer. He is survived by his wife and children.
Noah's wife has asked me to collect stories about Noah for the benefit of his children. Please leave a comment or share here: https://forms.gle/zxzsXkSdun1ahSqE6

2024-11-28 15:00:15

If you're running rack_timeout, you should probably also be enabling the term_on_timeout setting:
https://github.com/zombocom/rack-timeout/blob/main/doc/settings.md#term-on-timeout
It's basically impossible to have clean state after a thread kill. Need to recycle the process.
https://files.mastodon.social/media_attachments/files/113/561/174/949/705/848/original/de17e1419005fe37.png

2024-10-30 14:00:05

I'm going to make one of those latency charts you see for things like memory access speed, but do it for a Rails app context.
What things do you think would be on the chart?
i.e.
local variable access
querycache access
User.find(1)
A typical response
Elasticsearch

2024-09-16 22:32:59

- reply

@npub1pfe…e2dm
https://files.mastodon.social/media_attachments/files/113/149/606/394/990/889/original/cf6447426f1b5d6c.png

2024-08-28 14:00:04

> "Ruby isn't high performance enough for us"
> Their website downloads 7mb of JavaScript to display a blog post, makes 3 network requests in a row to hydrate data, re-layouts twice
https://files.mastodon.social/media_attachments/files/113/040/005/663/804/398/original/7b340b6441503983.jpeg

2024-08-28 10:57:23

Bwoaaahhh the new way Mastodon groups notifications is soooo much beeeettterrrr

2024-05-13 14:00:05

Almost everyone uses an exception tracker (e.g. Sentry) to track errors in production. These become bug tickets, sometimes even automatically. Bugs get fixed.
Do you have something like this for performance? If you don't, are you wondering why your engineers never fix perf?

2023-06-14 01:56:11

- reply

@npub1a4c…t2kf OKCupid seems like it was a really cool place to work back in the day. A lot of really cool people. @npub15rp…wudh

2023-05-23 14:02:47

Everyone should be using Spot instances for all web and background job workloads.
Your requests should be timing out after 60 seconds tops, and your background jobs should be resumable (or at least idempotent). Spot is not a problem. Take the 70% savings off your AWS bill.