CSEComputer Science & Engineering MapThe connected map of computing knowledge and engineering careers.
Show in graph
SYS

Software → System Design

SPOF

A component, dependency, or process whose failure can bring down an entire system or critical user flow.

SPOF diagram

Where it fits

SPOF stands for Single Point of Failure. It is a system-design risk: one server, database, queue, DNS dependency, team-owned process, or manual approval that can stop the whole service when it fails.

Mental model

A system is only as available as the weakest required dependency in its request path. If every successful request must pass through one component, that component can become a SPOF.

Example

A backend service may run behind many application servers, but still depend on one primary database instance. If that database has no replica or failover plan, the database is a Single Point of Failure.

How engineers reduce it

Use redundancy, failover, health checks, backups, multi-zone deployment, graceful degradation, and operational runbooks. The goal is not to duplicate everything blindly; it is to identify the dependencies whose failure breaks the most important flows.

Common mistakes

  • Treating redundancy as automatic resilience without testing failover.
  • Missing non-technical SPOFs such as one person owning a deployment key or manual recovery process.
  • Removing one SPOF while creating another, such as placing all traffic behind one unprotected proxy.

Availability, redundancy, failover, load balancing, bottleneck, reliability, and disaster recovery.