A defrag in time saves nine: Two NT disk optimizers

Ever notice how much more sprightly your computers are after you've reformatted the drive and reinstalled software? That's because the file fragmentation that occurs when you open, change and resave data files eventually slows the system down. And that effect can be especially noticeable on heavily used data and application servers.

Problem is that although defragmentation programs have been bundled into Microsoft Corp.'s Windows 95/98—as well as being available in a variety of third-party applications—Windows NT offers no built-in disk optimizer. And if you wanted to optimize an NT server, the only option available was Executive Software International Inc.'s Diskeeper. Now, Diskeeper has some competition from Symantec Corp.'s Norton Speed Disk 5.0. For the first time, the tried-and-true Norton product offers NT defragmentation and disk optimization in workstation and server versions.

The two companies have taken divergent approaches to NT defragmentation and optimization. NT's application programming interface for defragmentation was built by Executive Software, and it enables the company and other third-party software companies to hook defragmenters into the operating system for services. According to Executive Software, that's the only truly safe and Microsoft-approved way to do the job.

Norton Speed Disk, however, appears to sidestep the operating system and go straight to the BIOS and hardware for some operations. That enables the program to achieve things Diskeeper doesn't attempt—online defragmentation of the Master File Table (MFT) and paging files, for example. (Diskeeper can do those tasks at boot-up time, but who wants to have to reboot a server for defragmentation if it's not necessary?)

Online defragmentation of MFT and paging files may or may not be entirely safe, per Executive Software's dark hints. But Norton Speed Disk has been a longtime star in the disk optimization business, and parent company Symantec knows about mission-critical products. In any case, our testing was unable to reveal loss of data or other ill effects resulting from either utility.

Speed Disk additionally enables you to specify file placement optimization for better performance, such as putting your directories and most-used files at the start of the disk. Symantec also boasts that Speed Disk can defragment and optimize disks in a single pass no matter how fragmented they may be, while Diskeeper requires several reboots to defragment a disk for the first time if it's severely fragmented.

Diskeeper doesn't try to optimize disks by placing files in particular physical spots, on the argument that the uncertainties of disk usage make the benefits dubious. Regarding MFT and paging file fragmentation, the company offers a Frag Guard feature to protect those crucial system files from fragmentation, so there should be no frequent need to defragment them—online or off—once you get them in good working order.

For testing, we installed each program on multiple servers. Because Diskeeper advises users to perform multiple defragmentation passes, we ran each program twice before running the competing program to check whether the competitor was still able to find fragments to optimize. Then we ran one last pass with the first program to determine its ability to clean up after the other. we had to vary from this plan in testing Speed Disk, however, because the program performed so well in a single pass that subsequent passes were unnecessary.

Both programs can be installed across the network, a real boon to system administrators with many clients to manage. Diskeeper relies on Microsoft's extra-cost System Management Server (SMS) for network installs.

Speed Disk could use SMS or the included Norton System Center, which we found very capable. Both programs required clients to be rebooted after successful installation, and both programs support scheduling of fully automated background defragmentation for workstations over the network.

Speed Disk 5.0 was a breeze to install. The only snag was the fact that Speed Disk's installation program required certain files to be written to the C drive, which was not available on one dual-boot system. It's unlikely that servers actually in use will encounter this drawback, but it's conceivable that workstations might.

Single Pass Cleanup

The server on which Speed Disk was first tested was virtually a clean installation, so we figured it wouldn't be very fragmented. The program analyzed fragmentation at 33.5 percent, however, in an easy-to-read pie chart format. After a single pass, the analysis reported a fragmentation rate of 0.1 percent.

The next test involved running Speed Disk on a severely fragmented system that already had been defragmented by two passes of Diskeeper, leaving the drive still 27 percent fragmented, according to Speed Disk. That was reduced to 0.2 percent, again in a single Speed Disk pass.

We then used Norton System Center to schedule regular optimization for nodes on the network. That involved modifying a preset job for that type of operation. We could then select from various time intervals for optimization operations. All properly setup nodes were put on the same schedule.

Scheduling could also be done on each computer from within Speed Disk itself, with similar options except for the addition of threshold triggers. This is handy in case someone is running a notebook computer off the network and wants to launch a defrag operation locally, or for particular cases in which the global schedule is inappropriate.

Speed Disk's single passes were acknowledged by both programs as more effectively defragmenting the disks. It appeared that much of the difference between the two programs' effectiveness lay in Speed Disk's ability to defragment the MFT and paging files.

Moreover, Speed Disk's use of Norton System Center gives administrators a flexibility not offered by Diskeeper. If you don't need to manage Windows 98 or 95 nodes across the network, Speed Disk for Windows NT is an attractive solution.

Probably not coincidentally, Diskeeper did not install on the same system that balked Speed Disk. In Diskeeper's case, the problem began with a requirement for a more recent version of Microsoft Internet Explorer than the 4.01 version already present. The installation of Version 5.0, included on the Diskeeper distribution, failed because it couldn't find a certain system file (comctrl32.dll), but this error probably was related to the unavailable C drive.

On the severely fragmented server, Diskeeper's first run reduced fragmentation from a reported 2.02 average fragments per file (50.7 percent fragmentation, according to Speed Disk) to 1.35 fragments per file (32 percent, according to Speed Disk).

A second run reduced this number to 1.30 fragments per file (27 percent, according to Speed Disk).

After a pass by Speed Disk reduced the fragmentation to 0.2 percent, which showed up as a 1.01 fragment per file average in Diskeeper, Speed Disk's analysis showed only a temporary file remaining in four fragments, the MFT in three and the paging file in two. A third run of Diskeeper brought the statistics down to 0.1 percent, or 1.0 fragment per file—perfect, according to Diskeeper, although Speed Disk still showed the MFT and paging file as somewhat fragmented.

Trying the programs in reverse order on the relatively clean system didn't work because Diskeeper found nothing for itself to do after an initial run by Speed Disk.

Diskeeper optionally logged all kinds of disk statistics on both Windows NT and 95/98 systems. There also was a rudimentary advisory screen that popped up after we analyzed a given drive, but its value was dubious. It looked as though any detected fragmentation on a drive resulted in a statement that the drive was heavily fragmented, even when the program's separate analysis report said there was no need to defragment it.

Set and Forget

We were able to schedule defragmentation over the network flexibly and easily from within the Diskeeper server installation's "Set It and Forget It" window. Here, all the nodes on the network were selectable for individual or group settings. Diskeeper virtually enforces frequent defragmentation by providing scheduling options based only on hours and days of the week. You are allowed to pick your time range, though.

We thought Diskeeper's method of network scheduling was both easier and more flexible than Speed Disk for my small network's requirements. And Diskeeper works with Windows 95/98 nodes, a feat beyond the reach of Speed Disk.

So if you're running a mixed network and want central control over disk optimization, your choice may be made for you.

—Marshall is a free-lance writer who has been reviewing computer software for the past 10 years.


Diskeeper Server 5.0 for Windows 95/98/NT


Executive Software International Inc.

(818) 547-2050


Price and Availability: Available on the open market for $399 for Server and $75 for Workstation. For information, call (800) 829-6468.

Remarks: Diskeeper is a tried-and-true defragmentation program for Windows NT and also works well on Windows 95/98. Multiple passes of the program provide incremental defragmentation. You do, however, have to reboot your server if you want to defragment the Master File Table and paging file. Scheduling across the network is easy and flexible.


Norton Speed Disk for Windows NT Server 5.0


Symantec Corp.

(800) 441-7234


Price and Availability: Site licenses for Windows NT workstations are available at $33 per node for 100 workstations. Site licenses for Windows NT servers are available for $249 per server for 10 servers.

Remarks: Speed Disk's one-pass defragmentation is strikingly effective, and the program can optimize Master File Table and paging file online, a major advantage for departmental servers. The program's scheduling across the network is powerful, though not as easily modified as in Diskeeper. Speed Disk also optimizes disks by placing directories and most-used files, or user-specified files, at specified physical locations.


Stay Connected

FCW Update

Sign up for our newsletter.

I agree to this site's Privacy Policy.