Definition
Blast radius is the extent of a system that a failure, a change or a remediation action can affect: which workloads, namespaces, clusters, cloud resources, users or data are reachable from it. In operations it decides how much of an action to allow and how much verification it needs before and after.
Explanation
The term started in cloud security and architecture, where it describes how far a compromised credential or a failed availability zone reaches. In remediation it applies to the fix as much as to the fault. Restarting one pod has a small radius. Rolling back a deployment behind a shared load balancer affects every caller. Changing a security group or a cluster-wide config can affect everything. Blast radius is estimated from topology (what depends on the thing you are about to touch) and from the action type (read, restart, scale, edit, delete). It is bounded by limiting scope, by dry-running the change, by having a rollback, and by the permissions of whatever executes it.
How DataAgent approaches it
DataAgent maps the topology of workloads, cloud resources and their dependencies, which makes it possible to estimate the radius of a proposed fix before it runs. The guardrails set with the customer cover environment, scope of change and confidence, and the agent’s own permissions limit how far any action can reach.
Example
Two candidate fixes for the same OOMKilled pod. Fix A: raise the memory limit on that one deployment in one namespace. Its radius is that deployment’s pods, reversible with one edit. Fix B: change the namespace’s default memory limit. Its radius is every pod that restarts in that namespace from now on. A gate policy allows A automatically and holds B for approval.
Related terms
- Approval gates
- Guardrails
- Rollback
- Topology
- Dry run
- Levels of autonomy
FAQ
- How do you measure blast radius?
- Count what depends on the target: pods, services, callers, namespaces, data stores, cloud resources. A dependency map gives the count. Without one, assume the radius is larger than it looks.
- How do you reduce the blast radius of an automated fix?
- Prefer the narrowest action that addresses the cause, act on one instance before many, dry-run, keep a rollback, and limit the executor’s permissions to the scope it needs.
- Is blast radius the same as severity?
- No. Severity describes the impact of the incident. Blast radius describes the scope a failure or an action can reach. A low-severity fault can have a large radius if the proposed fix is cluster-wide.