single point of failure

One component whose failure stops the whole system working.

In more detail

Removing single points of failure usually means adding a redundant copy that can take over automatically. Finding them is a standard step when designing for reliability.

Examples

two web servers: doubling only the web tier helps, but the single database and power feed still dominate
Both sites use the same cloud account, identity provider and software build
A one-broker Kafka cluster is a single point of failure.
An automation orchestrator that every recovery step depends on is a single point of failure, and an attractive target: whoever controls it controls everything it can change.

Related terms

Taught in

References