Aws Data Migration Task have Latency Most of the Time

Question

I am working on a AWS DMS tasks and we have 7-8 replication process running but between all the replication task two tasks always have high latency issue I have checked our replication instance SPU uses is normal memory uses normal everything but still got latency don't know how? in the DMS tasks cloud watch matrix shows both source end point and target end point have latency but when we connect with our source team they always says our side is fine. So whats the problem with specifically those two replication tasks. Even if all the other replications are working fine with same replication instance same server. So is there any other methods in AWS to rectify this issue

I have tried stop resumed
I have tried restart the task

Answer 1

Latency can occur for variety of reasons which includes source engine, target engine (like s3 which consumes the incoming changes slow), slow network, number of changes being pushed in given task, and how the changes are being applied on target (Batch Apply vs Transactional), lobs, e.t.c

You need to look at Resource Usage - Cloud Watch Metrics, Task Parameters, and Task logs.

Please refer following links to understand the potential cause, and recommendations.

1. https://repost.aws/knowledge-center/dms-high-source-latency
2. https://repost.aws/knowledge-center/dms-high-target-latency