A financial services company is using AWS Database Migration Service (AWS DMS) to migrate its databases from on-premises to AWS. A database administrator is working on replicating a database to AWS…

While maintenance on a replication instance could cause temporary disruptions, it typically leads to a specific downtime event or restart, not a slowly increasing target latency, as the service is designed for high availability.

Modifying the source endpoint of an AWS DMS task generally requires restarting the task, which might cause a temporary pause or reset in replication but does not inherently lead to high and slowly increasing target latency during CDC.

"Loopback changes" typically refer to a scenario where replicated changes are sent back to the source or cause an infinite replication loop, which might lead to data duplication or unexpected behavior, but it's not a direct or primary cause of high target latency within DMS itself.

Without a primary key or suitable index on the target table, AWS DMS must perform a full table scan to locate the rows to update or delete during CDC, which is highly inefficient and causes target latency to increase significantly.

Resource bottlenecks, such as insufficient CPU, memory, or network I/O on the AWS DMS replication instance, can directly hinder its ability to process and apply change data efficiently, leading to a build-up of unapplied changes and thus increased target latency.