RabbitMQ Performance Improvements

on YouTube ↗53m

" summary

Presents concrete profiling and fixes: uses perftest, Prometheus/Grafana, microstate accounting, cprof/eprof and perf flame graphs to find CPU and mailbox hotspots; replaces expensive DB compilation for direct routing so direct-exchange lookups drop from ~10% to ~2.5% CPU and throughput improves 10–40%; fixes a dets/ETS bug with a one-line change that prevents 700k-message mailboxes and OOMs; ships direct exchange v2 behind a feature flag in 3.11 and advances quorum queues v2/classic-q v2 for lower latency and memory.

AI-written summary · about

" chapters (7)

1. 0:0001:23 - Intro & Agenda
2. 1:2406:16 - Erlang Programming Paradigms
3. 6:1713:11 - RabbitMQ Architecture
4. 13:1222:59 - Performance Tools
5. 23:0041:22 - Performance Improvements
6. 41:2347:14 - Outlook
7. 47:1552:58 - QnA

" related talks

• 36mDRb vs. RabbitMQ ShowdownDavy Stevenson · RubyConf 2012
• 26mBreaking Magical BarriersGerhard Lazu · Systems @Scale Spring 2021
• 32mBuilding distributed system for retail with RabbitMQMaryna Zhygadlo · RabbitMQ Summit 2021
• 50mPanel discussion: The new RabbitMQ StreamsKarl Nilsson, Arnaud Cogoluègnes, Diana Parra Corbacho, Gabor Olah, Ayanda Dude · RabbitMQ Summit 2021
• 59mDissecting the Rabbit RabbitMQ internal architectureAlvaro Videla · HighLoad++ 2013
• 59mDissecting the Rabbit RabbitMQ internal architectureAlvaro Videla · HighLoad++ 2013
• 54mKeynote: Bringing a Ferrari to a Horse RaceSophie deBenedetto, Steven Nunez · RabbitMQ Summit 2021
• 50m101 Ways to Cook RabbitMQ and the Pipeline ArchitecturePavel Filonov · HighLoad++ 2015