The backup server is pretty simple. It has two main components: a reader process (sybmultbuf) and a writer process (sybmultbuf) connected via shared memory. If you use the same backup server for the two instances, there should be no difference with respect to i/o between the two instances UNLESS there is a difference in the underlying disk.
- Check your os errorlog looking for file system i/o errors. When was the last time you ran a file system check?
- use iostat or equivalent to verify that there is an actual difference of i/o and not a perceived one. This is important.
- At what stage of the restore that you see the slowness?
