Five hundred workers call a downstream API every minute

A medium Pipeline Design interview practice problem on DataDriven. Write and execute real pipeline design code with instant grading.

Domain

Pipeline Design

Difficulty

medium

Problem

Five hundred workers call a downstream API every minute. A brief latency spike caused all 500 to time out at the same instant. Every worker retries on a fixed one-second sleep, so 500 retries arrive at the downstream simultaneously one second later. The retries themselves become the load. The section names this a thundering herd. Find the herd by replacing the fixed-delay retry with one whose name states both halves of the section's fix: a backoff schedule (wait longer between successive retries) and a jitter strategy (random offset that desynchronizes the wave).

Practice This Problem

Solve this Pipeline Design problem with real code execution. DataDriven runs your solution and grades it automatically.