Alistair Maclean's Web Site
The Geriatric Server Problem.
Back

Crash and/or Die?

In mid-1999, many companies froze spending on all IT purchases, that included servers and desktops. The Dot.com boom kept on going for a short time more but by the end of 2000 few companies were buying anything.

We are now just passed the mid point in 2002 with no immediate thawing of CFOs' wallets. We will probably go into the first or second quarter of 2003 with only a slow pick up in the general IT climate.

But what's happening behind the scenes?

Three and four year old servers are boat anchors - paperweights - door stops - Geriatric. They are remarkably unreliable and when they fail the cost of repair can be mind boggling. Machines this old have no intrinsic worth, they have probably been depreciated to nothing, and they have no resale value, assuming they are owned, not leased (don't you just wish it was a lease and you could force the issue of getting a new Server!) This means that the asset worth of the IT department is rapidly heading for zero (software has negligible value in accounting terms, never has, probably never will.)

It's hard ware, not infinite ware

Hard drives have MBTF (Mean Time Between Failure) numbers of between 10,000 and 100,000 hours (1 to 10 years of continous use), most drives tend to be of about 50,000 hours which gives them five year lives before a statistical half life catches up with them.

Power supplies are consumable items and fail, as are fans, keyboards, mice and monitors. Even the motherboard components don't live forever.

Over the next year many companies will have servers aging beyond their designed life span, yet there are clearly no plans to increase IT spending to account for their replacement price.

Most servers made in the last 2 or 3 years have had no ISA card slots. Older servers had these aplenty. Many tape backup systems used ISA bus cards. Replacement proprietary motherboards with ISA slots are expensive beasts, and nearly all servers have proprietary motherboards. Gone are the days when servers were desktops with big hard drives and no monitor.

A very few applications need hard drives of particular size, I have heard of older systems requiring 200Mb hard drives. These drives are simply no longer made, they have to be purchased used on the second hand market. They don't cost huge amounts, a couple of hundred dollars maybe, but the part has indeterminate life and its costs are horrible in comparison to buying a new 40Gb hard drive that runs faster and obviously stores far more - but of course won't fit the application. Many components are similarly compromised; video cards, network cards and SCSI controller cards are all now different than they were when you bought that server - many of the manufacturers no longer even exist.


The Geriatric Server is a computer that is mechanically old, its parts obsolete, its reliability suspect yet its usefulness is still considered appropriate for continued use.


Finding people that can still work on older operating systems has never been a good experience. The skills evaporate rapidly, but the cost goes up in terms of the hours needed to find old software and old replacement hardware.

The Replacement Crunch

After 9/11, the New York Metro area caused a major rush in server replacements. Some manufacturers are only now getting over the drain that had on their manufacturing. The load that might be placed on these same manufacturers in a few more years could be even more dramatic.

As enterprises put off the inevitable they also run the risk of obsolescence hurting them. They may get into a situation that the replacement hardware simply does not exist and a new server is the only option, however that server may require a newer, or different, operating system, which itself may no longer run the software that was put on the old box.

Obviously lack of profits or low asset use are put up as excuses for not spending, but soon that narrow thinking is going to dramatically skew the accounting numbers, as servers fail and replacements start to cost far more than the original purchase price. Servers failing and stopping business for days or weeks has a fairly profound effect on profitability, companies need to bear this tidbit in mind as they cull costs to save the bottom line.

© A. Maclean July 2002


[Opinions Index]
[page index]