MongoDB TTL indexes: free deletes, not free resources

I needed weekly report documents to disappear so an Atlas M0 cluster would not fill up. TTL indexes were the obvious move: set expireAfterSeconds on a date field and let MongoDB remove rows in the background.

They work well. The trap is thinking “TTL = no cost.”

There is no separate TTL fee, but:

• Documents still use disk until they are removed.
• Indexes still exist and are maintained on writes.
• Deletes are work the cluster does on a schedule — on a tiny tier, that matters if volume spikes.

Takeaway: TTL is great for temporary data you are okay losing on a timer. You still watch storage and write patterns. For anything mission-critical, I pair TTL with monitoring or a safety job that alerts before the cap.