We have extremely short retention policies for some jobs - effectively, I just need the latest copy of a VM stored as a backup (1-day retention if you will). However, if a job or backup of a VM fails, that causes VDP to destroy the only recovery point and the next backup is a full resulting in longer backup times, increased IO, etc. I tried to accomodate this by increasing the retention to 2 days for these jobs, but I can't go any more than that because of the data change rate and 2TB limit of the VDP appliance. That helped some, but still, jobs will fail and all recovery points will be expired before a new backup is taken. Is it possible to prevent VDP from deleting/expiring a recovery point if it's the only one for a vm?
if you are limiting recovery points due to space consideration of 2TB limit you might consider more than one vdp and spread the load across them to get more storage capacity and to allow for more recovery points. it could also provide a performance improvement if the destination is on different storage.