A Lambda function has Reserved Concurrency set to 100. Monitor concurrency for this function in CloudWatch.
Notify SNS when the maximum concurrency within one minute reaches 80. Notification only after requests are rejected does not meet the requirement.
Which metric, statistic, and threshold are most appropriate?
Alarm on the function's one-minute ConcurrentExecutions Maximum >= 80 and notify SNS as usage approaches 100.
Detailed explanation
Throttles counts already rejected invocations, rather than approaching the limit.
Throttles counts already rejected invocations, rather than approaching the limit.
Duration measures execution time, not a concurrency of 80.
Duration measures execution time, not a concurrency of 80.
Period invocation count differs from simultaneously active executions.
Period invocation count differs from simultaneously active executions.
This measures active concurrent executions against a threshold below the reserved limit.
This measures active concurrent executions against a threshold below the reserved limit.
Check official documentation
Compare the following points with the official documentation.
Compare Concurrency metrics, Throttles, and Reserved Concurrency in the linked documentation.
Official references
What to confirm
Distinguish the 80-of-100 early warning from a count of requests already rejected.
Key points
- ConcurrentExecutions
- Reserved Concurrency
- Throttles
Notes
- Environment: AWS SOA-C03 exam guide and the linked AWS documentation
- Reviewed against the linked AWS documentation on 2026-10-03.
Foundation review
Approaching versus rejected
Reserved Concurrency reserves and caps function concurrency. Use function-scoped ConcurrentExecutions with Maximum for early warning; Throttles counts rejected requests.
Warning does not guarantee prevention
High Invocations need not imply high concurrency when executions are short. A sudden burst during metric and notification delay can still cause throttling. Monitor rejection separately and plan a response.