Blast Radius
How much of a system a single failure or change can damage — you design to keep it small.
Blast radius is the scope of damage a single failure, bad deploy, or compromised component can cause. Good architecture works to shrink it, so one problem stays contained instead of taking down everything.
Worked example: a canary release limits a bad deploy’s blast radius to 5% of traffic; sharding by customer limits a corrupt shard to those customers; per-service credentials limit a leaked key to one service. Gotcha: the biggest blast radii are the shared things — a global config, a single database everything depends on, an admin credential, a shared cache — so reducing blast radius means partitioning, isolating (bulkheads), and least-privilege; a change with a large blast radius is exactly what needs a canary and a fast rollback.