Pattern 68 / LATENCY

Hedged Requests

Use this when Reducing tail latency from straggler instances.

Pressure: Reducing tail latency from straggler instances
Mechanism: Send a duplicate request after a delay and use whichever response returns first
Toll: Extra load can amplify incidents if hedging is not capped

Diagram for Hedged Requests — Architecture plate68

Executive brief

Hedged Requests fits when Reducing tail latency from straggler instances. Mechanism: Send a duplicate request after a delay and use whichever response returns first. Use it for Read-heavy services with variable latency. The toll: Extra load can amplify incidents if hedging is not capped.

Use when

Read-heavy services with variable latency.

Example

Use hedged requests when read-heavy services with variable latency..

Review framing

Describe the pressure first, then the mechanism, then the cost. That keeps the design grounded.

Same pressure family