One of the components most likely to fail -- and the one which often fails first -- in an HP 3000 is its disk drive. Consider the average age of disks attached to HP 3000s. Hewlett-Packard built the last HP 3000 and inserted onboard drives in that server one decade ago. Replacement and upgrade drives from HP, built after 2003, were for sale from HP for the 3000 through 2006. And there have always been drives purposed for one HP computer, but used for another. Those would be even newer devices.
All of the above devices are considered enterprise-grade. As the 3000 moves into its second decade of post-manufacture, owners will be looking for disk replacement strategies for the HP-branded servers. A virtualized unit, like the ones from Stromasys, have no such problems -- so long as their drives are of a high caliber.
But what is the caliber of a drive that is suitable for business enterprise use? A vendor of cloud-based computing argues that the failure rate of enterprise disks is actually a little worse than that measured for consumer-class drives. Through three years, one sort of drive might be replaced for another with little concern. It's possible, however, that years 4-10 are where the enterprise advantages emerge.
Jeff Kell, who's managed HP 3000s since the 1970s, as well as Linux and Unix servers more recently, said the promises of enterprise hardware for 3000s have never been guaranteed. That's especially true in an era where HP now won't warranty hardware of any sort attached to an HP 3000. But Kell added that pure math proves that drive failures will head upward as the size of the devices soar.
"I don't know overall if disks have gotten "better" or "worse" by themselves," he said. "But the sheer order of magnitude has certainly changed -- and simple math would show you the probability of error increases as the data density increases. Old disk drives only had to keep up with a few megabytes of data. Current ones may be a terabyte or more."
It turns out that the consumer drive failure rate does go up after three years, but all three of the first three years are pretty good. We have no data concerning enterprise drives older than two years, so we don’t know if they will also have an increase in failure rate. It could be that the vaunted reliability of enterprise drives kicks in after two years, but because we haven’t seen any of that reliability in the first two years, I’m skeptical.
Data loss is unacceptable for many customers, but for those who maintain their own drive farms, the blame can cost someone their job while it costs the company real money. An online solution costs money for the loser of data. The rest of the fallout is hypothetical.
Kell said that in a world where the HP 3000s often mirrored their data through RAID or Mirror/3000, 3000s still had failures of enterprise-caliber drives.
Our original 3000s had drive failures, sure. Then we had the days of Mirror/iX and having redundant drives. Then you get into the disk arrays; we've had both Nike and the VA arrays. Our "retired" N4000 had a 32-drive VA array, dual fiber channels, the whole nine yards. I think we replaced 4 drives over its lifetime of about 8-10 years. The current philosophies revolve around "expecting" failure, and keep on trucking, at various levels.