Managing RabbitMQ Usage at Scale for Bloomberg’s Trading Systems
summary
Bloomberg’s OMS uses RabbitMQ to handle 1500 avg / 40k peak msgs/s with a zero‑loss target by enforcing reject-publish policy and synchronous publisher confirms, and by using message priority and reply queues for UX feedback. A centralised service stores externalized JSON config artifacts that a client library reads to declare exchanges, queues, bindings and connection strings (enabling amqpprox swaps) so pipelines can add/remove steps or migrate connections without code changes. This system is live in production and used for topology visualization and large-scale sharding.
AI-written summary · about
chapters (8)
- 0:0005:40 - What is an OMS?
- 5:4108:17 - Our Score
- 8:1813:21 - The System
- 13:2221:37 - Why RabbitMQ
- 21:3828:51 - Externalized Application Config
- 28:5233:52 - Use Case
- 33:5335:43 - Looking Forward
- 35:4444:40 - QnA
related talks
32mBuilding distributed system for retail with RabbitMQ
36mDRb vs. RabbitMQ Showdown
16mTo queue or not to queue, that is the question.
50m101 Ways to Cook RabbitMQ and the Pipeline Architecture
26mBreaking Magical Barriers
18m"Kafka's Metamorphosis"
52mHow to Cook RabbitMQ Right — 8 Practical Cases
57mGame of Streams: How to Tame and Get the Most from Your Messaging Platforms