Conner CP3100

Photo: Conner CP3100, Red Hill.

Space: a major issue

Through the 1990s, you couldn't simply order a drive big enough to serve any rational purpose for the next four or five years the way you could in 2010, or even in 2001. It wasn't at all unusual to need 1 or 2 or even 4 gigabytes of storage, and the only way to get that was by going to a SCSI system, which was horribly expensive.

For example, our office 1.7GB Micropolis drive in 1993 cost close to $2000, plus another $800-odd for the necessary VESA controller card and cables. And that was pretty much the cheapest half-decent SCSI rig on the market.

Nor was it very practical to use several small, inexpensive IDE drives — most systems were limited to a single IDE controller, and even when twin controllers came into vogue around 1995, you immediately lost one of them to the CD-ROM drive. You could buy add-in IDE controller cards for PCI or VESA, but they were tricky damn things at the best of times, generally quite expensive, and notorious for having buggy drivers. The upshot was that more than two, or at very most three IDE drives simply wasn't practical.

But you had to store your data on the machine: none of the alternatives we take for granted today were viable. Networking was arcane, difficult, slow and expensive — and in any case, all it did was move the storage problem onto another machine — while external drives barely even existed, and certainly were not cheap enough or flexible enough for mainline storage. There were no CD burners, let alone DVD drives.

In the end, you wasted a lot of time trying to find space on your hard drives, or else you went to an enterprise-standard SCSI system, which gave you effectively unlimited storage space with much bigger drives, and the ability to add as many of them as you cared to pay for. But SCSI cost a fortune.

As for large IDE drives, they were a long time coming. The technology to make the drive itself was readily available (SCSI drives had offered very large capacities for some years past, albeit at high cost) but there was no simple way to make that space available. The eventual answer, Logical Block Addressing or LBA, needed cooperative action from drive manufacturers and mainboard BIOS vendors, and for reasons hard to fathom LBA mainboard BIOS did not reach the market until around 1996.

Micropolis (and, from memory, one or two other manufacturers) offered peculiar conversions of big SCSI drives to IDE from as early as 1993, but the full capacity could only be accessed via a boot sector device driver such as Disk Manager, better known to the technicians of the day as Disk Mangler. In theory this worked quite well. In practice, the tiniest accidental change or error in the boot sector could and did cause catastrophic loss of the entire contents of the drive. In an era when boot sector viruses were at their peak of popularity, this was madness.

Finally, there was the Seagate method: safe, effective and remarkable in its simplicity. Most Seagate drives in the 500MB to 1GB size range included a dual drive emulation jumper. Set that jumper and (as if by magic) so far as the computer was concerned you had two 500GB drives, a master and a slave, and you could access the whole 1GB that way. Obviously it only worked to full effect with drives no larger than 1GB (though Seagate provided it nevertheless on the 1.28GB ST31270A) and less obviously it couldn't increase your total capacity because the single physical drive occupied both the master and slave positions on the ATA controller. Just the same, it was a handy stop-gap and helped tide things over until ATA-2 and LBA arrived in 1996.