This question pops up often, and there's rarely a good answer given: Why is this one particular backup so slow?
How about a diagnostic tool for troubleshooting slow backups?
Random thoughts follow:
-Have the tool do a speed test with the agent. Is it network related?
-Have the tool get more information from the agent. Is the agent getting enough CPU time? Is the agent getting files from disk fast enough (i.e. Defragmentation, SAN issues)
-Help parse out what parts of a backup job are taking the most time. (With work, I can get this from old job logs now)
-Is the tape drive/library keeping up with the data coming from the agent? Any obvious signs of SCSI errors talking to the drive?
-Is the server getting enough CPU time? Is the server getting bogged down with too many active jobs? (It could happen, in theory...)
-Have the tool scan the backup selection list, and report the number of files and median file size. Report how this will impact expected speeds.
-I've heard that the agent compresses data before giving it to the server. If so, report how much CPU is spent doing this, and how much compression we're getting.
-How much time is spend mounting/dismounting tapes?
-Look for & recommend obvious performance improvements. Pre-scan turned on? Server NIC set to auto-negotiate? Agent NIC set to auto? Backing up pagefile.sys or hiberfil.sys?
I'm envisioning a tool that doesn't nessecarily pinpoint my performance problem, but at least tell me where it's not, and roughly what part of the system it is. Find the bottleneck, so I can focus on a fix.
