We tested the 26 tape-backup subsystems using a hierarchical file structure with five directories and three subdirectory levels. We evenly distributed about 100 MB of test data. File sizes in each subdirectory are randomly distributed and range from 3072 bytes to 3072 KB, with compression ratios ranging from 10 percent to over 85 percent.
We used two Compaq Deskpro 66Ms (486/66 EISA), each with 16 MB of RAM and an IDE hard drive. We installed NetWare 3.12 on one system to act as a file server. We installed MS-DOS 6.2, Windows 3.1, and Windows NT Workstation 3.5 on the other system to act as a workstation.
If the vendor supplied a SCSI card, we used it; otherwise, we used an Adaptec AHA-2740 SCSI controller. For those units supporting parallel-port backup, we used the parallel port only if the vendor did not supply a SCSI adapter and the
bundled software was intended for parallel-port use.
TEST METHODOLOGY
Our test procedure consisted of two distinct phases. In the first phase, we tested every entry using the vendor-supplied hardware and software. If the supplied software was an NLM (NetWare loadable module), we attached the tape drive to the file server and created the file structure on the server volume. After we backed up the data in overwrite mode, we restored it to the server. If the vendor-supplied software was a workstation product, we attached the tape subsystem to the workstation and created the file structure on the workstation's local disk.
If no software was supplied, we used Cheyenne Software's ARCsolo 3.02. We chose ARCsolo for its ubiquity in the marketplace and its use of a nonproprietary database (Novell's Btrieve) for tape library information. However, the same traits that make ARCsolo almost universally compatible also limit it: Advanced features such as record keeping and error-correction
control are not supported.
The backup tests ran once with hardware compression disabled and once with it enabled. After the initial backup and restore, we backed up the new directory and restored it down a different path. For each drive, we repeated the tests until the variation between runs was reduced to under 5 percent. To reduce backup and restore time, we disabled such features as backup and restore verification and NetWare bindery file backup/restore options.
The second phase of testing used the same data tree locally on the NT machine, with NT drivers and backup software. We used an Adaptec AHA-2740 SCSI controller for the Windows NT tests, whether the vendor supplied interface hardware or not. Because the NT tests all used the same software and SCSI controllers, differences in performance can be attributed to the drives themselves.
FEATURES AND EASE OF USE
To evaluate ease of use, we examined the documentation and the use of LED indicators for tape operation a
nd fault isolation. We considered ease of use during both setup and configuration.
The features score reflects the range of supported software and hardware platforms, controls, and other options.
Our overall ratings combine performance scores with usability and features scores. Higher scores are better.
Contributors
Tadesse W. Giorgis, Project Manager/NSTL, has tested NOSes for NSTL for over five years. He holds a Ph.D. in fiber and polymer science from North Carolina State University.
John McDonough, Technical Editor/NSTL, has been writing for high-tech publications for the past five years. He can be reached on the Internet at editors@nstl.com.
Other testers were Samir Abzakh, Vidya Navarathna, and Michele Guy.
l.com; at NSTL, Inc., Plymouth Corporate Center, Plymouth Meeting, PA 19462; or at (610) 941-9600. Contact BYTE via the Internet or BIX at editors@bix.com or at (603) 924-2643.
How We Tested
