Hi Osborne,
I have a couple of questions/suggestions:
1. The expectation would be that the SetCopy job (which is what I gather you're running) would be as slow as the slowest drive. With that said, what kind of speeds are you getting when a job is done straight to the LTO3 library? I imagine it would be quite faster than straight to the LTO1, but I need to double check.
2. If you are willing to experiment, can we try replacing one of the tape drives with a B2D device to try and isolate the bottleneck? In other words, let's try a backup with a significant amount of data to a backup to disk folder, and duplicate that to the LTO1 library. Once we find that speed, let's do the same test, only using the LTO3 library.
3. Since you are now running 11d, you can also use the SCSI tracer to capture some of the READ/WRITE traffic to see what the I/O throughput looks like at the SCSI level. Here is a technote on how to utilize it:
http://support.veritas.com/docs/285630
I'm also checking with our QA department right now to see if this is a normal scenario. I'm not sure if we'll release any hard number evidence, but at least we could say whether or not you should expect a decrease in performance using this feature. I'll let you know what I find, but please reply back with the answers to the suggestions/questions above.
Thanks.