kaplan71 (8/25/2014)
Hello --I have been monitoring the amount of time it takes for a Full backup to complete. The e-mail notification shown below provides an example of the current state of affairs:
JOB RUN:'Backup Plan.Full Backup' was run on 8/23/2014 at 10:00:00 PM
DURATION:26 hours, 7 minutes, 44 seconds
STATUS: Succeeded
MESSAGES:The job succeeded. The Job was invoked by Schedule 9 (Backup Plan.Subplan_1). The last step to run was step 1 (Full Backup).
The amount of time that it has taken to complete such a backup has grown over the course of several months from just under an hour to just over a day. There are a total of twenty databases that are part of the maintenance plan, and they range in size from fifty megabytes to 133 gigabytes.
The mdf files are located on a drive connected to one disk array, and the bak files that are generated are on a drive that is on another disk array. The two disk arrays are connected to a private local network that utilizes a 10-gigabyte switch.
As part of the troubleshooting process, I have replaced several disks that were reporting a relatively high amount of I/O errors in the array that hosts the bak files. The disks in the array contain the mdf files did report any issues. The Full Backup log file did not have entries that indicated a problem with the actual job.
Beyond that, what other steps could I take to isolate the problem? For example, the Windows Performance Monitor has SQL counters available. Which of these would be useful in this case?
Quick thought, this sounds like a serious bottleneck somewhere. Have you checked the health status of the two file systems, fragmentation, collision rates on the switches etc.?
😎