Server Disk Management in a Windows Enviornment

The following is a summary of disk optimization vs. defragmentation:

• Potential for Gain. Due to the mechanics of head movement and disk rotation, the maximum possible gain from optimization is only one or two milliseconds per seek, but only in the best-case scenario. With significant fragmentation, however, as much as 15 minutes can be added to boot time and files can take 15 seconds or longer to open, so the potential for gain from defragmentation is many orders of magnitude greater than anything that can be offered through optimization.

• Overhead. Apart from the fact that relatively minor returns are to be reaped, optimization also seems to consume more system overhead than a regular defragmentation utility. I tested this briefly using the Windows Task Manager, and the optimization programs run more slowly than regular defragmentation programs. After all, the process involves the analysis and repositioning of every single file on the disk to achieve what is believed to be an optimum file pattern. According to NSTL, the cost in resource overhead to accomplish such a task can result in a net loss to performance. I tried one vendor's optimization product and was discouraged with the length of time it took to reposition everything on my disk, far longer than a defragmentation run. Although future runs are supposed to take far less time, I never experienced much gain in the procedure and found it somewhat expensive in terms of overhead.

• Safety. The New Technology File System (NTFS) is a sophisticated, high-performance I/O subsystem that requires extremely close synchronization between the memory manager, the cache manager, and the file system. That is why reliable defragmenters are programmed to move as few files as possible. They work in close coordination with the applications programming interfaces (APIs) specially written to allow defragmentation on Windows NT and Windows 2000, moving only those files that absolutely must be moved. As a result, a minimal amount of overhead is expended, while gains are maximized and risks are minimized (about one percent of files are moved during defragmentation by the best utilities). Moving 100 percent of the files, on the other hand, provides 100 times the opportunity for error, greatly increasing the dangers of file corruption.

One optimization product I checked out was plagued with problems and has since faded as a marketing presence and is now included only within a utility suite. Users have reported problems with PowerPoint files after optimization, experiencing graphics degradation or files that will not open. Others have suffered from system slows with Outlook and other applications. To avoid the potential for sluggish performance or file corruption, the safest solution appears to be to shut down all other applications before running optimization software — perhaps possible for the individual consumer, but unworkable for most enterprises. It is worth noting, though, that Raxco's optimization product does not have a reputation for file corruption. The first run does take overlong, in my experience, but that company has managed to avoid the corruption problems associated with other optimization vendors.