Due to the cyclical efficiency gains in ingest scripts coupled with broad and aging hardware platforms we are experiencing degenerative processing capability that results in break scenarios for rapid batch ingests (commonly known as 409 errors). This is causing operation adversity that can be alleviated by moving from inline to managed datastreams.
This is a continuation of https://jira.duraspace.org/browse/ISLANDORA-675 .
Instead of just targeting metadata datastreams we should target all inline datastreams.
The specific trouble is that Fedora runs into write locks with the FOXML and 409s under heavy load. Heavy load is determined both by how much work Fedora is being asked to do (both on a FOXML and system level) and the hardware it has available to do it. This most often has catastrophic consequences for rapid batch ingests but in some cases impacts front end single object ingests.