Is there a similar performance impact, or any impact at all, when restoring deduplicated data?
If you are restoring
a single file, odds are you aren't going to see a performance impact there. It's when you are talking about doing full system restores that it really starts getting more complicated. It really depends on how the index is constructed and how the data is laid out on the back-end disk. As long as the index can remain in memory, it can continue to process and find data as it's reconstructed.
The problem that begins to emerge is two-fold. First off, if you have backups going on at the same time you are trying to do restores, there is contention for resources. Second, if you have a large amount of deduplicated data, the index will be too big to be kept in memory. So, some of the index is cached off to disk. Swapping the index out from memory to disk and back again can dramatically increase backup time.
Check out the entire Data Deduplication FAQ.
This was first published in December 2007