It really depends on your requirements, hence why we discussed the approach with 3 different criteria. If precision is important to you, then yes, you’re right, it’s not a good fit for you.

But equally, there are use cases where scalability is of paramount importance and precision is of little consequence. One of my clients, for instance, needs to schedule millions of tasks over vast spreads of time. They need the task to be executed reliably at some point after the threshold (but not before). Precision is not important to them, but the approach needs to scale to support many millions of open, scheduled tasks. So, DynamoDB TTL is actually a pretty good fit for their requirements and is a very cost effective solution too.