MENTAL MODEL · M4494

Thundering Herd Problem

BusinessHigh supportManagement

Included

Updated 2026-08-02

Loading revision record…

INTRODUCTION

English translation pending.

CORE DEFINITION

In concurrent computing, the thundering herd problem describes what happens when a large number of processes blocked on the same event are all awakened by one notification, although only one can proceed. The wasted wakeups consume CPU and lock contention, and the losers must sleep again. The cure is selective wakeup: condition variables, edge-triggered notification, or randomized jitter that spreads the retries over time.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

- Warn about synchronized wakeups: treat any broadcast signal that rouses every waiter as a congestion risk - Spread the load: add random jitter so simultaneous demand fans out over time - Narrow the wakeup: notify one capable worker instead of the whole pool whenever possible

anchor

Anchor fast decisions

All waiters on one event are awakened by the same broadcast, but only a few can acquire the resource. The rest burn CPU, reacquire locks, and fall back asleep. Broadcast wakeup without selection turns a single event into a burst of contention, so throughput falls as the number of waiters rises.

MINIMUM ACTION

In progress 0/1

Practice this model in one real situation:

Replace blanket wakeups with precise notification, such as condition variables or edge-triggered events. 2. Limit the number of workers allowed to handle any one event at a time. 3. Add randomized jitter to retry and polling intervals so demand spreads over time. 4. Measure wakeup storms and requeue costs before assuming more parallelism helps.

Check to track your progress (stored locally)

Learning progress0%

account_treeGenealogyexpand_more

menu_bookReferencesexpand_more

Source support: Explicit

link
en.wikipedia.orghttps://en.wikipedia.org/wiki/Thundering_herd_problemZH · Explicit
verified

RELATED MODELS