Question

A financial services company runs a critical order-processing Lambda function that is invoked asynchronously from an EventBridge rule. During a marketing spike, the DevOps team notices that many events are being lost and the function's Throttles metric is spiking. Investigation shows the account's unreserved concurrency pool was exhausted by an unrelated batch-processing Lambda function that scaled up aggressively. The team must ensure the order-processing function always has capacity available AND that any events it cannot process immediately are retained for later reprocessing, without losing data. Which combination of actions BEST meets these requirements?

Accepted Answer

Configure reserved concurrency on the order-processing function to guarantee capacity, and configure an on-failure destination (or dead-letter queue) on the function's asynchronous invocation settings to capture events that exhaust retries.

Answer

Configure provisioned concurrency on the batch-processing function so it stops consuming the unreserved pool, and enable EventBridge rule retries with a longer maximum event age.

Answer

Increase the account-level concurrency limit via a service quota request and rely on EventBridge's built-in 24-hour retry to redeliver throttled events.

Answer

Set reserved concurrency on the batch-processing function to zero during the spike, and configure an EventBridge archive on the bus to replay lost events later.