The JPA source uses a background garbage collection mechanism to clean up any large values that are no longer needed. However, some use cases that don't change much data will not need such a process, and so for them this background process is just unnecessary overhead.