A comparison of the different IR workflows with an opinionated default and tests against previous incidents.
See:
Reconstruct timeline
Look for deviations from existing models
Look for broader patterns in limiting or reducing failure
3a. bulkheading failure (by feature or traffic percentage)
3b. advertising failure behaviours to clients
3c. degrade to cached content (hard for write clients)
n. propagate knowledge (or mark as evidence for tooling?)
n+n. pattern match post mortems; look for common terms or so.
A comparison of the different IR workflows with an opinionated default and tests against previous incidents.
See:
tests