In Nashville last week, an erroneous impression was communicated by one of the speakers at the CDW event suggesting that Data Domain — purveyors of storage arrays with built-in de-duplication — wanted folks to de-dupe their data then send it to tape. We managed to include Data Domain in a Q&A on the event to set the record straight. If companies are using tape, DD wants you to “re-inflate” your data before streaming it to tape media, lest you (1) incur a serious delay on data restore from tape when you need it (i.e., following a disaster) and/or (2) create a need to field another DD box at your recovery site. Sending de-duped data to your disaster recovery backup tapes makes no sense and will introduce a substantial time penalty into restore operations.
Having said that, a rep from the company told me off line that they will be shortly introducing and NDMP backup strategy for the de-dupe data stored on their box. They already offer the capability to replicate (mirror) data between two or more DD units, but the fellow told me that some customers “having achieved a measure of confidence in the de-duplication technology we offer” are considering going straight to tape with their de-duped data. Hmm.
It’s one thing to have confidence and to want to reduce the amount of data to be backed up by squeezing the shite out of it, but it still flies in the face of restore timeframes which are already painfully expanded by the current wave of “encrypt everything” thinking among the compliance-concerned.