Load balancing: an in-depth study to scale @ 80K TPS
summary
Covers practical methods to scale load balancing to ~80K TPS by comparing DNS, Layer 3/4 and Layer 7 approaches and showing HAProxy examples and monitoring metrics. Walks through hardware vs software routing options including LVS (history and modes: NAT, direct routing, tunneling, RP filter), interrupt/CPU-affinity handling, and HA with Keepalived and Consul, plus recommended things to monitor.
AI-written summary · about
chapters (16)
- 0:00Start
- 1:18Evaluation criteria
- 2:21DNS Based
- 3:31Layer 3/4 Load Balancing
- 3:52Haproxy Monitoring
- 5:29Benchmarking Env
- 8:25Issues with Haproxy L4
- 9:11Layer 7 load balancing
- 11:09Routing L3/4
- 12:33Types of routing
- 14:36Software Router: LVS
- 15:41LVS Issues
- 16:05LVS: CPU Affinity
- 20:13LVS: RP Filter
- 23:36LVS HealthCheck
- 25:15LVS IPIP Debugging
related talks
30mKeeping the Balance: Internet-Scale Loadbalancing Demystified
46mKeeping the Balance: Load Balancing Demystified
24mReliability Is Not Just About Balancing Loads
40mGood, Bad, and Ugly - the Art of Load Balancing at Scale
18mDeveloping a Custom Load Balancer to serve traffic at scale
38mThe Untapped Power of Services - L7 Load Balancing Without a Service Mesh
35mMake Load-Balancing Great Again with !
44msmart balancing with nginx+lua