[OPLINTECH] Computer Replacement Schedule

[Nathan Eady]

"Paris Robertson" <robertpa at oplin.org> writes:

> My Board is looking into adopting a computer replacement schedule, and
> I was wondering if any of you had some sample policies out there.
> They are looking at roughly a 5 year rotation period.  Any information
> you can give would be helpful.

If you're getting a new policy drafted, try to get it in writing that no
new non-replacement equipment can ever be added (either via purchase OR
via donation) unless the ongoing budget is augmented to allow for
replacing it as it ages.  Otherwise, they'll keep trying to add nice
shiny NEW computers you previously didn't have, without increasing the
budget for replacements.  At least, they keep trying to do that to me.

Also, I have our different stations divided into "tiers".  I've found
this categorization a useful mechanism to track how replacements are
handled.

The first tier consists of systems that have substantial performance
requirements, so they are replaced only with new systems.  This is,
unfortunately, the largest category.  If you find that that's the case
for you as well, you may wish to get the policy written mainly around
the first-tier systems.

The second tier consists of systems whose performance does theoretically
matter, but in practice it doesn't matter as much, e.g., because nothing
they do really exercises the capabilities of a modern computer.  Also,
some systems are in tier two rather than tier three to avoid the worst
reliability problems, even though performance is unlikely to be a big
issue (e.g., the mail server and the firewall are in tier 2).  Tier 2
also includes ready "loaner" spares for a couple of the tier-one
categories (e.g., a spare patron internet station, a spare staff ILS
workstation).  I used to put systems mainly used for browsing the web in
the second tier.  Unfortunately, with the recent proliferation of
absolutely ridiculous quantities of gratuitous client-side scripting in
the last handful of years, it has now become necessary to place most web
browsing stations in tier 1.  This has roughly doubled our number of
tier-one systems and almost makes me want to add a "tier zero" for
things that _really_ need to perform, but I digress.  In any event tier
two gets hand-me-downs from tier one, so even the _newest_ tier-two
system is older than any of the tier-one systems.

The third tier gets whatever is left when the second tier is done.  Most
of our tier-three systems are dedicated web-catalog kiosks, running
Debian stable with a custom X session that makes sure there's always a
browser window open.  (Our web catalog doesn't seem to require much
performance on the client side, so it does okay even with quite old
hardware.  YMMV.)  If you still have dedicated word-processing stations,
they can go in tier 3.  Certain kinds of low-impact servers (e.g., print
servers) can be tier 3.  That legacy system you keep around for testing
stuff like how your website looks in ancient web browsers from the days
of yore (*cough* IE6) can be tier 3.

I try very hard not to let first-tier systems be much more than five
years old.  Second-tier systems cat get to be seven or eight years old.
(Third-tier systems can be older; we currently have one vintage 2002 HP
Pavilion that is still in service.  Use 'em till they're dead.)

Getting timely upgrades to happen here has entailed creating a special
presentation for the board featuring a color-coded graph, with the
oldest systems in the most alarming colors, showing a side-by-side
comparison of what the future will look like if we replace 2 computers
per year versus what it will look like if we alternate 3/4/3/4.  (When
the bright red edges over into neon magenta, people are starting to pace
the room while waiting for the computer to do something.)  Geek that I
am, I sort of thought it would be rather obvious that the number of
systems you replace per year multiplied by the age you are willing to
let systems get to before you replace them cannot ever be less than the
total number of systems you have in service, but apparently this is too
much math for some people.  Hence, color-coded visual aids, and hence my
suggestion about getting mandatory ongoing replacement funding for all
new systems written into policy.  The need for up-to-date computer
equipment is never so dire that it can't be made worse by increasing the
number of systems that need replaced without increasing the ability to
replace them.

If wishes were horses, I'd consider trying to ensure that all first-tier
systems are still within three-year warranty range (because systems in
that age range are significantly less prone to failure, not because the
warranties themselves are particularly valuable).  Realistically,
however, our budget isn't going to go the distance for that.  YMMV.

-- 
Nathan Eady
Galion Public Library