When your server is under stress, it is possible that the NTFS log file may not be flushed to disk. If the NTFS log file becomes full, NTFS file operations are suspended until the log file is completely flushed to disk.
To determine if your reduced performance is caused by the NTFS log file filling up, use Performance Logs and Alerts to inspect the activity of the Current Disk Queue Length counter for the desired PhysicalDisk object. If the Current Disk Queue Length counter is constantly high, with intermittent drops to one (exactly 1), the NTFS log file is full.
NOTE: When the Current Disk Queue Length counter is 1, you cannot write to disk.
The preferred method of resolving this problem is to upgrade your hardware, to include bus speed, disk controllers, disks, and possibly RAM. Use striped volumes over multiple physical drives to increase throughput.
As a temporary workaround, you can increase the size of the NTFS log file:
1. Open a CMD prompt.
2. Determine the current NTFS log file size by running CHKDSK <Drive:> /L.
3. Increase the NTFS log file size by typing:
CHKDSK <Drive:> /L:newsize /F
NOTE: You may have to schedule the change for the next restart.