halcanites wrote:I am running DP 8.10 on win server 2012 R2. Is there a way I can see the throughput of the backup job while it is running? Also, is there a way I can see what file is being backed up while the job is running?
Regarding throughput, it becomes second nature to "feel" the throughput from watching the Monitor GUI if you're "doing DP" for some years. It really escapes me why a display of the instantaneous transfer rate was never implemented, given everybody and their dog repeatedly ask for it. When the feel is somehow wrong, I help myself by watching things like the Task Manager network tab or atop(1) on Linux (to see the incoming data rate on the MA hosts), and Resource Monitor, Perfmon, Process Explorer and again atop(1) to watch source disk saturation, single core CPU saturation and such. To see the file being backed up, if the DA in question runs on Windows (I presumed so from your sparse input) you could start Resource Monitor (from Task manager performance tab), filter yourself on the vbda process in question and then closely watch the "Disk" pane on the Overview tab. On the CPU tab in the "Associated Handles" pane you will get even more info on the filtered process, specifically open files it isn't actively doing I/O with. When you're already here, you may find a reason for the bad throughput - have a close look at Disk "Highest Active Time" (in the Overview and Disk Panes).
Whether the throughput you see is about right, nobody can tell without more data. 12TB could be a dozen ideally unfragmented large files, or it could be 1 billion really small ones in a file system with metadata looking like it took several buck shot. Assuming 100 IOPS per spindle (typical estimate for mediocre disks, its mostly 80..200 from worst to best), a worst case 4KiB random access pattern needs 12000000000000/(4096*100) seconds to read 12TB, that is 340 days! A lots-of-small-files traversal is close to pure random access, so any throughput beyond 6MB/s is actually not that bad and harder to achieve than most people think. You need multiple spindles for that. Your volume K took 20 days to pull 6TB? Thats approx. 3.5MB/s. That requires 800 IOPS at worst case randomness, which would mean some 6 to 8 spindles to achieve even that on spinning rust. Or some less, because the worst case isn't usually what you hit. So it may be completely normal. Watch your source disk saturation.
HTH,
Andre.
