[Click Here for MD+F Text Effects 3D!]
[Previous]
Hard Drive Optimization- by Sam Henwrich
[Next]

Summary: Round 1: It's taking 10 minutes to load the operating system, 2 minutes to start your mail client, it's time to think about defragmenting the contents of your hard drive. Which utility is best when your daily activity stresses even HPFS 's ability to keep files together in single continuous chunks.

A hard drive does not truly store information like a filing cabinet, that is, you can't just push files down to occupy the empty space left behind by ones you've thrown away. If a new file must be saved that is larger than any single continuous block of free space, then it will be broken into two or more chunks. The file system transparently keeps track of where the chunks of each file are kept, so to you, it still appears to be a single and unbroken file. The FAT filesystem that comes with DOS is terrible at this, it doesn't even bother looking for areas of large enough free space first. HPFS is better, but it's not immune.

The solution is to periodically run a defragmentation utility. These take a couple of hours to run, depending on the size of your hard drive and severity of fragmentation. They work by re-ordering the files so that they span the minimum number of separate chunks or "extents." FAT defragmentors have to do the most work, while their HPFS counterparts have it easier. With HPFS, the filesystem itself merely has to be given a second chance to re-arrange files into unfragmented blocks. But any FAT utility must explicitly do all the calculations itself and override the operating system.

Since OS/2 can both be installed to and natively manage both filesystems, Gammatech and Graham Utilities both have the ability to defragment drives formatted in either. However, neither package does this all within a single program. You run a different program for each file system supported, often with a different user interface and different behaviors. For example, Gammatech has a PM (GUI) defragmentor for HPFS drives, but it's FAT equivalent runs in a text-mode window. Graham, which is the most consistent between utilities, will defragment a drive visually (showing a map of the drive), but while it offers a compressed view of the drive's contents in the FAT utility, it does not in the HPFS equivalent.

User Interfaces

Gammatech utilites wins the ease-of-use category for its HPFS optimizer, which comes in both a PM and full-screen version. You can select the drive to work on with a click, set it's priority and click to go. What's unusual, but possibly useful is a built-in scheduler that will run the optimization process at a set time of day or period of time. We found that the command-line equivalent of the utility could be easily run unattended by a CRON or other kind of scheduling program too, however, which may be more desirable.

Graham Utilities makes up for this, however, with a visual defragmenting utility that shows a map of the hard drive's contents, changing the display as it goes through the optimization process. Gammatech Utilities features a visual optimizer, but only for FAT partitions, and without a PM version.

Methods of work

Each utility suite uses a different method when it comes to defragmenting HPFS drives.

Graham Utilities lets you specify exact files or file masks, letting you defragment the contents of only one directory at a time, or the whole drive. The actual mechanics it uses underneath are simple: It lets OS/2 do all the work and merely identifies the files that are fragmented, moving the file so the operating system can re-seat it in a continuous slot. It's sort of the hard drive equivalent of giving the carpet a good shake and letting the furniture fall back in a more agreeable pattern.

Gammatech is quite different. For a start, it takes the whole drive in one go, even though it does offer the ability to deselect files it has chosen for optimization before it actually begins the process. Secondly, it seems to take a more direct route to defragmenting the files, rather than letting OS/2 do the work as with Graham's defragger. We have not yet confirmed this as fact yet, but our comparative testing led us to believe so.

Job Completeness

Where the two suites differ even wider is how complete they consider the job to be. It was with disappointment that we found Graham utilities often left the drive with barely any serious defragmentation performed at all. After running it once we followed it up with Gammatech's optimizer, which reported over one and a half thousand files left in a fragmented state.

This was easy to obverve as the program ran too. Graham's HPFS defragmentation utility frequently reported that "Free space is too fragmented to allow defragmentation". Perhaps only a handful of files were ever optimized in the whole, 2 hour long process. Graham Utilities' program did not attempt to defragment the free space.

Gammatech Utilities' HPFS optimizer was a different story, when we could get it to run without freezing up the whole system (as it did a number of times during our testing. We've been unable to determine if this was because of a conflict with other software, or a flaw within the program itself). It does a more thorough job, either defragmenting the free space or making multiple passes until the files sift down into optimized state. When it's done, it's pretty much done.

Conclusions

Both competing utilities need enough free space on the drive being optimized to temporarily store the largest file to be defragmented. But the less free space there is, the more likely it is to be fragmented itself, and Graham's HPFS program won't be able to do a complete job. Gammatech's optimizer was the only one that could manage in such a constrained environment.
Gammatech Pass. Good PM interface and a optimizing routine that doesn't stop until the job's done. Big problem with locking up the system, however
Graham Fail. Nice visual optimizer, but performs ineffectively, especially when free space is also fragmented.

[Previous]
 [Index]
 [Feedback]
 [Next]
Copyright © 1998 - Falcon Networking ISSN 1203-5696
November 16, 1998