Solving High-load Load Balancing Problems
summary
Describes high-load load-balancing problems encountered at Qiwi and various techniques to increase fault tolerance. Presents qiwi-thrift-pool, a load-balancing library used for microservice architectures, and explains how to use it in Java, Kotlin, and Scala projects.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:13Round Robin limits
- 2:39Least connections issues
- 4:06Weight balancing challenges
- 5:55PID instability
- 7:05Thread pool drawbacks
- 9:33Queue overflow effects
- 11:23Cascading blocking
- 13:09Microservices overhead
- 14:57Binary protocol benefits
- 15:40Custom load balancers
- 16:47Discovery service need
- 17:53Sharding and key balancing
- 20:01Consistent hashing
- 21:48Pool size pitfalls
- 22:32Summary & Q&A
related talks
46mKeeping the Balance: Load Balancing Demystified
53mHTTP Traffic Load Balancing
19mLoad is not what you should balance: Introducing Prequal
35mRethink About Service Mesh Load Balancing
01h 01mgRPC client-side load balancing — why, what, how and WTF?
30mKeeping the Balance: Internet-Scale Loadbalancing Demystified
17mQDSR: Accelerating Layer-7 Load Balancing by Direct Server Return with QUIC
33mCloud Native Load Balancing From Scratch