Establish what is happening
Record burst shape and affected actions
Capture arrival rate, concurrency, duration, route mix, request sizes, response statuses, latency, and the user actions waiting on the system. Mark when the burst began and ended, plus whether traffic was expected or synthetic. Keep load evidence bounded to an approved nonproduction or controlled environment.
Compare steady and burst timelines
Use matched route mix and request shapes at steady load and during a bounded burst. Compare admission wait, application time, database and connection usage, queue depth, errors, timeouts, and recovery. Identify the first metric that changes with concurrency rather than treating all degradation as one failure.
Check whether recovery is clean
Observe the period after arrival rate returns to normal. Record lingering queues, rejected work, elevated errors, stale state, duplicate retries, and time to return to the steady pattern. A burst that finishes quickly but leaves backlog or incorrect work belongs in the reliability scope too.
What to carry forward
Return a burst scope statement naming load shape, first degrading resource or interval, affected actions, errors, and recovery behavior. Route the next work to capacity, admission, queueing, or correctness. Stop when a controlled comparison identifies the next measurable boundary without assuming a scaling mechanism.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow