blast radius

Definitions

1. The scope of impact when a failure, misconfiguration, or security breach occurs — how many users, services, or systems are affected. Good architecture minimizes blast radius through isolation boundaries: separate accounts, regions, availability zones, and microservice boundaries. The question 'what's the blast radius?' should precede every production change.

   In plain English: How much damage a failure can cause — good system design limits the blast radius so that when something breaks, it only affects a small part of the system.

   Example: The blast radius of that database migration was the entire customer base because we ran it on the primary instead of testing on a replica first.

Origin Story

The military metaphor that teaches engineers to contain their mistakes

**Blast radius** was borrowed from military and demolition terminology into software engineering during the rise of microservices and SRE culture in the 2010s. In its original context, it measures how far destruction spreads from an explosion.

In software, blast radius describes how much of a system is affected when something goes wrong. A monolith has maximum blast radius -- one bug can take down everything. Microservices, feature flags, and canary deploys all aim to reduce blast radius.

The concept is central to chaos engineering and deployment strategy. Netflix's deployment pipeline deliberately limits blast radius by rolling changes to one region at a time. AWS uses 'cells' and 'shuffle sharding' to isolate failures. The goal: when (not if) something breaks, minimize the collateral damage.

Coined by: SRE/DevOps community (borrowed from military terminology)

Context: 2010s, industry-wide adoption

Fun fact: Amazon's 2012 'Wheel of Blame' incident retrospective revealed that most outages weren't caused by bad code but by deployments hitting too many servers at once. This led to their cell-based architecture that limits the blast radius of any single deployment.

Related Terms

• Chaos Engineering
• Circuit Breaker
• error budget
• ADR
• Container
• Decentralization
• Deployment
• Distributed Systems
• Event
• Game Loop
• Instruction Set
• Message Passing
• Offline-First
• Replication
• Safety-Critical
• Sharding
• Supervisor
• System Design
• Transaction
• Greenfield
• Environment Variable
• Chaos Monkey
• Von Neumann Bottleneck
• Local-First Software
• Canary Analysis
• Dependency Graph
• Cache Invalidation
• Second System Effect
• Dark Debt
• Error Boundary
• Error Code
• Error Correction
• Error Handling
• Error Recovery
• Error Type
• Frameworks vs Libraries
• Fullstack Framework
• GUI Framework
• Helm Template
• Master-Slave
• Message Bus
• Multi-Tenant
• Pull Model
• Request Pipeline
• Toolchain
• Unidirectional Data Flow
• Vagrant
• Virtual Device
• Core Service