In a business situation, where people's time costs money,
spending 2 days (16hrs at, lets say $50/hr == $800) to screw
around with marginal old hardware for a fileserver isn't worthwhile.

When I can go to local computer store and get last years' good
machine for $400 (with a free printer!) and toss BSD onto it
(took 1 hr last night) and be running a machine that will NOT
die in the next year and will be running with reasonably current
hardware and perhaps a warrantee.

For similar $$$, I can build a much better machine with top quality
components (RAM, Power and cooling are key to the life of a computer).

And I can spend the rest of that time (1.5 days) dealing with
customers, doing the other system admin work I need to do, whatever.

I've spent more time dicking around with customers who want to
keep that Sparc 5 up as their server, paying me $5000 do deal
with its problems, when they could pay a fraction of that and
get a better, faster machine.

Sometime you have to pull the plug (says the guy with several
Sparcs, an Alpha, two NeXTs and an SGI Indy at home:).


Quoting Wojciech Puchar (wojtek@chylonia.3miasto.net):
...
> thinking that way - you should say support for other platforms that
> PeeCees are waste of time. fortunately it's not.
> 
> but buying 900Mhz machine for routing is a waste. even if it costs a
> little (350$ isn't a little - depends on country where you leave).
> 
> there are enough people (mostly windows users but not only) saying "why
> think, better buy new computer as it's cheap".
> it's definitely the wrong way. Yes - time is money, but unused organs
> stops working, including brain.
> 
> There are tons of almost free (or free) hardware in 386/486 class, enough
> for routers, X-terminals (PCI ones, lots of nices GFX cards are almost
> free too) etc.

Free until you work around broken video, time spent to make it as
fast as your network etc.  Acquisition of hardware is almost always
a TINY percentage of the amount spent running it.