Low latency Change Stream for Global Cluster in Atlas
our event-driven applications need to publish events to Kafka triggered by Change Stream feature. It works perfect in replicaset MongoDB cluster.
However, after migrating to Global Cluster in Atlas, the Change Stream cannot keep low latency because of ordering reason among shards.
The latency may go up to 20 seconds when for a single change event.
It would be nice if application can receive Change Stream from the single shard only (not care about ordering among shards) to prevent latency.
The idea is to pass "location" options when starting the Change Stream cursor.
3
votes
Vincent C
shared this idea