Building blocks & how to use them: MongooseIM case study
summary
MongooseIM shows concrete results: tests hit 2.5M concurrent connections on a large AWS instance with delivery <100 ms, and configuration was collapsed into one huge map stored in persistent_term yielding measurable microbench gains. A custom XML parser runs about 20x faster than OTP’s, a segmented ETS cache avoids 500x duplicate DB fetch storms, and contributions land in OTP and fast_tls adding TLS 1.3, SCRAM and channel binding support for real deployments.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:20Who am I
- 2:52What do I do
- 4:06Performance
- 6:54How does it work
- 7:14Configuration
- 9:35Architecture
- 10:49Pools
- 12:22Telemetry
- 16:16Config
- 17:56XML parser
- 21:08syslogger
- 22:10stateless libraries
- 24:24segmented cache
- 27:21pull requests
- 29:18HTTP tools
- 31:52Debugging tools
related talks
40mBuilding blocks and How to Use Them: A MongooseIM Case Study
28mMongooseIM - The Right Tool for Scalable Messaging
40mDeploying node.js at scale
01h 14mEfficient and reliable microservices
40mOpen Riak how we gained from leaning on the BEAM
24mGoing CPU Bound
41mIdioms for building fault-tolerant applications with Elixir
30mMonolith-ifying perfectly good microservices