failure domain
A group of components that can fail together because they share something, such as a rack, power supply or zone.
In more detail
Capacity is planned so the system still carries peak after the largest failure domain is lost, usually one availability zone. A container or pod is not a failure domain; if all copies of a service sit on one node or rack, that becomes the real one.
Related terms
- availability zone — One of several isolated groups of data centres inside a cloud region, with separate power and networking so a failure in one does not take down the others.
- blast radius — How much of a system is affected when something fails or has to be running for a check to happen.
- anti-affinity — A scheduling rule that keeps copies of the same service on different machines, racks or zones.
- headroom — Capacity kept free above the expected peak, so the system survives deployments, spikes and failures.
Taught in
- Back Of The Envelope Estimation: Traffic And Throughput — Scalable System Design & Distributed Architecture
- Capacity Planning As A Design Input — Scalable System Design & Distributed Architecture