Hi Ann,
I'm assuming this is V8 since you are seeing replication issues. One thing I would suggest is to examine your heatmap and see if you have this delivery trying to kick off around the same time when there is a large volume of workflows being used. If so, then I would try spreading out the workflows to minimize the number of replication calls occurring at the same time.
Additionally, there has been a bug reported for this process which has been fixed and is pending a release. NEO-73955 is a fix to the code that helps optimize the replication queue handling process, so you will want to keep an eye out for that in upcoming release notes.
If spreading out the workflows does not alleviate your issue, this could be a resource constraint at the Snowflake warehouse level, which Campaign R&D would need to address and require you to submit a ticket with Campaign Support.
Regards,
Craig