I am in the process of evaluating VTL options for TSM. We are interested in Data Deduplication but in my experience the vendors all have a very difficult time discussing this topic in a TSM environment.

When one does Full and Differential backups it is easy to discuss this topic. When one does Incremental forever backups it seems to be a big mystery. As we all know, in this model the only data being backed up is data that has changed. Since deduplication benifits from like data patterns, the fact that we are only getting changed data would seem to throw a wrench into this.

I am curious what other TSM users have seen with Data Deduplication from vendors like DataDomain, Dilligent, Sepaton, or others. I assume the 20:1, 50:1 marketing numbers are nonsense.