A client with a small server was running out of disk space. It is Dell PowerEdge 830 with 2 GB RAM running under Windows 2003 Standard.
The machine came from Dell with three 72GB Seagate (Cheetah 10K.7, 10K RPM, ST373207LW Ultra U320) SCSI drives which Windows saw as a 15GB C: drive and 125 GB D: drive. (The drives are attached to a PERC 4/SC RAID controller.)
Instead of replacing the existing SCSI (RAID) array I decided to just add a second array with three more Seagate 300GB SCSI drives. (I can't believe I did not log their model number!) That went fine. Windows now showed a new 600 GB drive. It partitioned and formatted uneventfully.
I then copied everything in the D: drive to the new one, deleted the old D: drive, and made the new 300 GB volume the "new D:".
Shortly after that, the client reported that when they called up some (graphic) documents (like faxes) in their application they experienced random delays of 5, 10, sometimes 15 seconds.
They found that the delays go away for about a day or so if they do a defragmentation. I personally think defragmentation is one of the most overrated operations I know and it cannot be a direct factor in an operation with only 10 or 12 terminals which does NOT expand many files. Mainly, they create NEW ones.
So, I think the defragmentation cure is actually a "misidentification". I think Defrag is doing something else (it reports 0 fragmentation when launched) that alliviates the problem but I cannot figure out what.
Basically, when they are looking at a screen, they select some fax or image they want to see and it sometimes "stalls" the application for 5, 10, or even 15 seconds in that they must wait that long to see the image. (The application happens to be eClinical Works, a program to run a paperless medical clinic.)
It smells like perhaps swapping activity in the (dedicated) Server but . . . What's going on here??