MongoDB and DynamoDB
Latest stable Based on Leader release 0.4.0
Coordinate with atomic document updates or conditional key-value writes, then account for TTL and clock behavior.
MongoDB
Section titled “MongoDB”The MongoDB backend uses atomic findOneAndUpdate ownership changes and a TTL index for cleanup. It offers blocking and suspend variants and supports the shared single-leader extension path. TTL cleanup is not an exact deadline, so correctness relies on the acquisition predicate, not deletion timing.
DynamoDB
Section titled “DynamoDB”The preview DynamoDB backend uses conditional writes and logical TTL. Design capacity for synchronized contender bursts and monitor throttling. DynamoDB TTL deletion is asynchronous; the stored expiry condition, not physical row removal, determines acquisition.
When to use
Section titled “When to use”Use these backends when the application already treats the store as an operational dependency. Test with the real service or a faithful emulator, including clock skew, conditional conflicts, throttling, and retry ambiguity. A client timeout may leave the write outcome unknown.