Wrong answer on Lambda reserved provisioning

[fernando-6]

This answer is wrong. Yes, the part about reserved provisioning prevent on Lambda function from taking too many requests is correct. However, nothing in the question says it’s ok that some of the S3 uploads messages will be potentially lost if they were throttled. In fact the question specifically says the object replication does not need to happen “in-real-time”, but it doesn’t say it doesn’t need to happen at all. The “correct” answer just says we can monitor, but doesn’t say what can be done. The queue approach is actually the classic scenario for using a queue. We can accumulate some events, and eventually catch up. That’s what queue are made for. Sorry, but this is just plain wrong.

[Carlo-TutorialsDojo]

Hello fernando-6,

Thanks for your feedback.

I understand where you’re coming from, but the ‘lost messages due to throttling’ is its own separate problem. In the given question, we’re solely concerned with preventing Lambda functions from hogging concurrency from each other. The usage of DLQ for storing failed requests is actually a valid solution to these kinds of problems so you can retry or debug later on. But as I’ve said, that issue is no longer within the scope of what’s being asked in the scenario.

The reasoning behind the option in the red box is that it ‘requires more work.’ However, the scenario didn’t specify a need for the least amount of work. We acknowledge this oversight on our end and will work to improve it.

Regards,

Carlo @ Tutorials Dojo