MENTAL MODEL · M1740

Rolling Update

TechnicalHigh supportSoftware Engineering

Included

Version 1.0.0 · Updated 2026-07-30

CORE DEFINITION

Instead of updating all servers at once, update them in batches (e.g., 10% at a time). If problems are found during the process, you can stop at any time, and the impact is limited.

SCAFFOLDING EFFECT

psychology

Reduce cognitive load

Smooth transition. When implementing new policies, do not switch all staff suddenly. First pilot in one department (rolling), and if no problems, then roll out to the next department. This spreads the pain of reform over time.

anchor

Anchor fast decisions

Software deployment strategy that replaces old instances with new versions in batches (rather than all at once), keeping the service uninterrupted during the update, and allowing pause and rollback on anomalies.

MINIMUM ACTION

In progress 0/4

Practice this model in one real situation:

Group instances into batches, take old versions offline and bring new versions online batch by batch
perform health checks and monitoring after each batch
stop and rollback on anomalies
control batch size to balance speed and risk.

Check to track your progress (stored locally)

Learning progress0%

account_treeGenealogyexpand_more

menu_bookReferencesexpand_more

Source support: Explicit

link
zh.wikipedia.orghttps://zh.wikipedia.org/wiki/%E6%BB%BE%E5%8B%95%E7%99%BC%E8%A1%8CZH · Explicit
verified

RELATED MODELS