IBM's mini computerslack thereof 1868

It's all I've had experience with. One pair of VAXen with a pile of spindles on them were running a Semiconductor Fab, they had to work 24x7x365. Failure was simply not economically viable.

Wrong atbreastude. It's more of a case of they never anyone any grief, they were taken for granted. Think of the stories of -8s and -11s that kept chugging away so long people forgot they existed.

No, it should happen because the bottleneck was the rate at which the heads could write data to the spinning dust. There are options such as solid-state storage that can help here, but they would not have satisfied the auditing requirements (as I understood it).

To give you an idea, the SCSI busses, controllers, memory and CPU were running at sub-15% capacity at year-end. FWIW it the OS had bugger all to do with most of the I-O on those disks, Sybase was looking after that, and it did do a damn good job. quicker in every dept barely managed an overnight batch in realtime. :(

The algorithms were nigh on optimal given that particular workload, we looked into it. The disks were within 80% of the best at the time, and they were plenty quick enough to get the job done (sub 15% cap on a year-end is *very* low).

HSCs seemed like overkill to me as well, but in practice they seemed to keep those semi-conductor fab VAXen running for a decade.

Regardless, the Alphas I saw were throwing Terabytes around no problem, 300 interactive users at peak. IMO those boxes were plenty good enough, perhaps SGI might have been able to offer something better, but I doubt it at the time, AFAIR O2K wasn't there yet. Even if the O2K + XFS were there, it would have made f.all difference to Sybase. :)

Cheers, Rupert