Comparison of Backup Products

Backup of data is probably the most neglected aspect of system administration in small businesses and home offices.
PerfectBACKUP+ 6.2
  • Manufacturer: Merlin Software Technologies, Inc.

  • E-mail:

  • URL:

  • Price: $69 US, $89 US for box and treeware

  • Manufacturer: Enhanced Technologies Software, Inc.

  • E-mail:

  • URL:

  • Price: $245 US

  • Manufacturer: Knox Software Corp.

  • E-mail:

  • URL:

  • Price: Free or variable; Quote

Quick Restore
  • Manufacturer: Workstation Solutions

  • E-mail:

  • URL:

  • Price: Quote

  • Reviewer: Charles Curley

Backup of data is probably the most neglected aspect of system administration in small businesses and home offices. Yet it can be critical. If your entire business is riding on the contents of your hard drive, and it goes belly-up, you are out of business. Worse, you may be legally liable. For example, you may be required to meet your payroll within a certain time after the end of the pay period.

For these reasons, what you spend on backup, including software and tapes, is a form of insurance. To give one example, I administer a home office with five computers (two Linux, two Windows 95, and one that multiboots between Linux, Windows NT and Windows 95), and two users. The network is standard 10-base T. I have two tape drives (a Conner/Seagate 4GB Travan and an HP DDS 3 DAT drive) and routinely use two of the products I discuss here.

Four backup programs for Linux are reviewed below: PerfectBACKUP+, Arkeia, BRU, and Quick Restore. The first three are available on the Web. Arkeia comes packaged with Red Hat, Mandrake and SuSE distributions of Linux. Quick Restore is available on CD only.


This is the only completely “no charge” program in the lot, and coming into such a program, I was not expecting much. I was surprised, for example, to see support for tape changers, something I would not expect to see on a low-end backup program. However, the program qualifies as “nagware” by giving you a “please register” window when you start up the program (see Figure 1).

Figure 1. The Main Window of PerfectBACKUP+

I was very pleased to see that it detected and correctly identified both of my SCSI tape drives. Whether this facility extends to other interfaces I cannot say. The two drives were not associated with the correct device files in /dev, so I hand edited that. A quibble indeed, compared to the setup problems I have had with other tape backup software.

One thing that is very nice about PerfectBACKUP+ is that the device selection menu identifies the drives by their names as provided by the device. That is very useful; it saves having to remember device names.

Figure 2. Selecting a Destination Drive for a PerfectBACKUP+

PerfectBACKUP+ provides some easy-to-use minimal acceptance tests, a facility I'd like to see on other backup software. One quibble here is that I'd like to see the software test with larger data sets to better exercise the tape drive. The data set is unfortunately so small as to make the data rate result meaningless.

Defining a backup set (or package, in PB's terminology) is easy: go down the Backup pull-down menu and define each characteristic in turn. Then select the Save As Package menu entry, and give it a name. You can then recall the package and run it all you want. I was able to define a small test backup set and make the backup easily.

Unfortunately, the restoration had problems. I got a cryptic message about “This backup was made from / and you are about to restore it to /test. If your backup was RELATIVE you might not want to do this.” Relative to what? Grepping the documentation, I found no explanation of what the term “relative” means. It then asked if I wanted to change the working directory. If I refused, it gave me a cryptic error message. If I accepted, and retyped the working directory I had previously selected, it made the restore where I had asked for it.

This is the kind of glitch that indicates inadequate testing. I can accept that sort of glitch in the GUI, as long as I can work around it. But I am left with the question: what else was not adequately tested? Also, PerfectBACKUP+ left defunct processes on my test computer. This is not encouraging.

Verification is easy. In addition, after the fact—weeks or months—you can pop a tape into a drive and verify it. You can verify the contents of the tape against the original files, or you can run a CRC checksum verification against the tape alone. This is useful for verifying the reliability of your tape drive: remounting a tape is a good test of the repeatability of head positioning, a major casualty of wear in lower-end tape drives like QIC tapes.

Unfortunately, I saw no way to verify a backup as part of the backup sequence. This is something I have been accustomed to seeing in PC backup software since Colorado Memory System's software provided it in the late 1980s.

There are two sets of documentation. One is in HTML, and you read it via Netscape. (What happens if you don't have Netscape?) The documentation screen shots indicate an xterm interface with mouse support; however, the provided interface is a full X GUI. The other documentation is available from pull-down menus on the GUI. It has no search function, and uses a fixed font which is vanishingly small if you have a high-resolution monitor. The two sets of documentation contain different material. Neither one documents the command-line interface, which is a pity.

I must reluctantly conclude that PerfectBACKUP+ is not yet ready for prime time. It needs serious testing in the GUI and a lot of cleanup in the documentation. Because it left defunct processes on my system during my tests, I don't trust it. PerfectBACKUP+ has some excellent features, like identifying drives by name. I hope Merlin fixes its problems and makes a more solid version available.