There can sometimes be issues with meeting SLAs while using DPM within tight constraints. The problem is that while consistency checking is essential, it kills other scheduled synchornizations while it runs.
"Synchronization with consistency check, also referred to as “a consistency check,” is the process by which DPM checks for and corrects inconsistencies between a protected volume and its replica. As part of the synchronization process, a consistency check performs block-by-block verification to ensure that all the data on the replica is consistent with the protected data. This process is slower than incremental synchronization because all the data on the replica is compared rather than just applying the data changes to the replica."
Lets say your SLA is something very low such as having the ability to restore to the last 30 minutes. We set up a 30 minute synchronization in DPM, which in itself isn't bad (though it seems to be more commonly set around 3-4 hours). The issue comes when the consistency check runs, which could take for example 1.5-2 hours. This would result in at least 3 failed synchronizations, thus breaking the SLA.
The only methods I am aware of to somewhat work around this would be to modify the SLA, or manually schedule consistency checks.