- Use a replication alarm threshold more like the snapshot threshold, since it runs just as frequently (40)
- Catch when too many snapshots are in progress and log it, vs. triggering a ticket (40)
- Add CreateSnapshot permission explicitly to the role used by Lambda (40)
- Adjust the CFN resources to be sure the trigger is setup correctly (40)