I have similar backup infrastructure as yours, and we do Opt-Dup for some backup replication from one site to another.
If it's Opt-Dup, the duplication of backup images are processed by the storage appliances (such as DD). There are two benefits: 1. Your NBU servers are kind of 'freed" from the hard work of copying the backup images. 2. The two DD's talk to each other and only copy the unique chunks of blocks that the target DD doesn't already have. Therefore it requires less network bandwidth utilization and is "Optimized".
You will see such line in the duplication job's status/details:
09/25/2011 21:57:10 - Info Duplicate (pid=14193) Initiating optimized duplication from @aaaaG to @aaaaC
Here you can see the phrase "optimized duplication" and you can see the copying of backup image from one DD disk pool to another DD disk pool.
While the duplication job is running, you can log on the DD's console and run these commands to check the Opt-Dup job performance and statistics:
# ost opt-dup show performance
# ost opt-dup show stats
If it's not Opt-Dup, the duplication of backup images are processed by the NBU Media server. In such case your source and destination DD do not talk to each other, and all the backup image replication is reconstructed by DD (from deduped/compressed to original) and read by NBU Media server and write to another DD (then dedupe/compress again). You won't see the "optimized duplication" phrase in the details of that duplication job, and you would only see resource request/grant and bpdm or bptm processes running on the NBU Media server for that duplication job.
