How do we build high‑load microservices
summary
A concrete production stack for high‑load microservices is presented: about 200 service types across five self‑hosted datacenters handling ~500 concurrent games and ~80k logs/sec. The system uses Kafka for streaming, Microsoft Orleans for stateful actors, Consul + Fabio for discovery/routing, Aerospike as cache, InfluxDB for metrics, Gatling for load tests and property‑based tests for correctness; the talk shows the built architecture, observed failure modes, measured performance numbers, and which fixes proved effective.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 10:10Tech stack
- 21:50Functional programming in C#
- 33:29Property-based testing
- 44:30Integration testing
- 50:11Q&A
related talks
33mPros/Cons of Highly Scalable Microservice Systems
44mLong-lived microservices
54mBuilding a High-Load-Capable REST API
49mChallenges in Implementing Microservices
48mChallenges in Implementing MicroServices
28mMicroliths: how to avoid traps in your stack and your culture
07h 30mHighLoad++ 2018 Live Stream. November 8, Congress Hall
28mMicroservices – Boyan Dimitrov