Growth from zero to 15,000 messages per second: painful and instructive
summary
TIMCONNECT processes bank SMS messages and attempts to deliver them as push notifications. The talk describes how a small team multiplied throughput by 15× in two months with minimal code changes, using high-load design patterns and trial-and-error, and presents the mistakes encountered and generalizable lessons for scaling similar systems.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:31Business challenge
- 2:57Initial architecture
- 8:03Load testing
- 12:41Failed optimizations
- 16:41Sharding solution
- 21:06Worker scaling
- 22:52Final testing
- 23:53Disk I/O bottleneck
- 24:20Key lessons
- 27:40Q&A
related talks
34mHigh-Throughput Push Notification Delivery
49mThat's 'Billion' with a 'B': Scaling to the Next Level
39m400 Billion Messages a Day with Kafka
01h 13mMessage Processing Failed! But What’s the Root Cause?
35mScaling Erlang cluster to 10,000 nodes
28mConversational big data
40mReal Time Performance at Massive Scale
41mLessons Learned Implementing a Resilient, Distributed, Real-Time Architecture