Decoder. plain-English AI glossary

Failover

● Core

Automatically switching to a backup system when the primary one fails — ideally without users noticing.

Think of it like

An elevator with a backup power system that kicks in if the main power cuts.

Example

A request gets routed to server A; server A crashes; the load balancer automatically sends it to server B.

How it actually works

Failover can be automatic (health checks detect failure and redirect) or manual. Automatic failover is complex — you need to detect failure quickly (false positives trigger unnecessary switches), coordinate state, and maintain data consistency. DNS failover is slow; load-balancer-based failover is fast.

For product teams

Good failover is invisible to users and crucial for uptime guarantees.

For engineers

Implement health checks carefully; test failover scenarios; use local retry logic.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome