Index

DevOps · Load balancing: tail latency

The load balancer

Every algorithm looks identical on the benchmark. Which one is still standing when two requests in a hundred cost sixty times the rest?

Eight identical machines share one stream of requests, and something in front has to decide who takes the next one. The bars show where the work actually landed, and beside them sit the wait a caller suffers and the deepest queue on any one machine. Make a few requests far more expensive and the four choices stop looking alike.

Board: Load Balancing & DNS →

Eight load bars, two comparison tables and a DNS readout. On a narrow screen the panels stack and the tables scroll sideways.

try this Drag How much more the expensive requests cost to the far end, then switch tabs and read p99 under What the caller waited.

Included with a plan

This lab is included with a plan. The first three lessons of every track are free, with their boards and videos. A plan unlocks every lesson, board, video, lab and paper for one price.

View plans →Already a member? Sign in

see also The Tail at Scale in Papers.