Defragmentation utilities are difficult to assess because you don't have a common measuring stick by which to compare performance across products. Therefore, to judge a partition's condition and the effectiveness with which a product defragments that partition, you must rely on the reports that each product generates. However, reporting discrepancies are common among the products. For example, each defragmentation utility that I tested can express the level of fragmentation as a percentage, but the formula each product uses to calculate that percentage is different. On an identical partition, Perfect Disk, Speed Disk, and Diskeeper reported the percentage of fragmentation as 7 percent, 20 percent, and 61 percent, respectively.
To initially determine fragmentation levels, I arbitrarily chose Speed Disk's percentage calculation. To provide detailed and useful numbers, I studied each product's reports, filtered out problematic vendor-specific formulas, and extracted the most relevant data. The result is the following group of tables, which are a composite of all the products' reports. Each table describes a particular partition in raw numbers, followed by the results of each product's attempt to defragment the partition.
I began with approximately identical C and D partitions for each server—I mirrored all of them from the same source disk. The C partitions had approximately the same amount of data with differing amounts of fragmentation. Partitions D1 through D3 had increasing amounts of data with a consistent level of fragmentation (about 50 percent). D4 contained a large Microsoft SQL Server database and log file.