Apache RocketMQ - Trillion Messaging in Practice
summary
Claims handling trillions of messages during Alibaba’s Double 11 (2016) festival and exposes what actually enabled it: a log-structured commit store with per-consumer queues, delayed delivery via a hierarchical timing wheel, transaction messages with two-phase commit semantics, SQL92-like server filtering, MQTT and batch send support, and HA coordinated by controller + ZooKeeper. Performance tuning details include JVM choices like G1 GC, Linux page-cache and page-fault tuning, and operational metrics showing sub-10ms latencies at massive throughput.
AI-written summary · about
chapters (26)
- 0:00About RocketMQ
- 0:54Welcome
- 1:45About Alibaba
- 3:40History
- 5:55Application Integration
- 9:38Domain Model
- 14:27Features
- 17:45Guardian Message
- 18:09Account Transfer
- 19:35Commercial Distribution
- 19:53MQ Features
- 20:38Track Messages
- 21:06View Messages
- 21:26MQTT
- 22:01Architecture
- 22:34Include Me
- 22:52Cluster Listing
- 23:22Consumer Status
- 23:39Query Message
- 24:02Query Message Details
- 25:02Chemical Performance Tuning
- 28:39Development Line
- 28:54Ecology Overview
- 29:21Open Messaging
- 31:32Articles
- 31:55Tools and Machine
related talks
46mReal-world Examples of Message Queue Usage: Insights from RabbitMQ
40mBuilding Scalable Resilient IoT Messaging
24mAdvanced RabbitMQ Strategies: A case study
22mLessons Learned from Scaling RabbitMQ in a High-Growth (...)
35mNo Rocket Scientist Required - Developing +14,000 Message p/second IOT Systems
41mForget about HTTP
32mRabbitMQ in Real Life - Tips and Tricks
59mAdding 1.21 Gigawatts to Applications with RabbitMQ