-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

At some point hitherto, Michael O'Donnell hath spake thusly:
> because I don't see how owning a new (versus
> used) RAID controller saves you if it fails.
> If you can get a replacement controller under
> warranty why can't you get a replacement
> controller by other means?  Anybody here who

The point is, whether or not it's a good value to buy a used part
vs. a new part depends on many variables, including the specific
hardware, the user's need for that hardware, the vendor and their
warranty policy, the relative costs, and the purchaser's aversity to
risk.  So it's not a clear-cut argument, either way.

I think it's largely a risk management argument.  If you decide you
want to buy cheap, used RAID cards, and it fails, you need to buy
another cheap, used one.  How many failures does it take for it to
become more expensive than buying a new card, which is replaced for
free under manufacturer's warranty?  Whether or not the used card is
more likely to fail than a new unit of the same item is a function of
things you probably can't know, mainly how the hardware was treated
prior to it arriving on your doorstep.

Therefore, it becomes difficult to assess the risk involved in buying
used cards (or, as a measure of risk, how much it will cost over some
period of time to buy used RAID cards).  If you go with used cards,
and they typically cost you $100/card, and the MTBF is 1000 hours,
then every 10,000 hours you're looking at $1000 in used cards.
This ignores the fact that the used card will have some of those hours
already used up, in all likelihood.

[Note, I'm not saying that is a reasonable figure, I'm only picking
the number out of a hat to use as an example.]

OTOH, if a new part has the same MTBF (1000 hrs) but has a lifetime
manufacturer warranty, and it costs you $400, then every 10,000 hrs
you're looking at $0 plus your original $400.   That makes it a better
deal.  In case it's not clear what I mean, the costs for used and new
cards in my example are:

used:  100x
new:   0x + 400

where x is the number of failures.

It's also worth noting that some hardware which has a lifetime
warranty may have a *transferrable* warranty, so if you buy it used,
you don't lose out if it dies.  Other hardware may not have a
transferrable warranty.  You may or may not need to provide proof that
you're eligible for said warranty, which you may or may not be able to
do.

If the $400 for a new card is a lot of money to you (or the
organization you're buying for), and you *need* the part in question
for some important use (i.e. you're not just buying it to play around
with SCSI hardware), then depending on the terms of the warranty, it
may well be worth it to buy the new part, even though it costs more up
front, because it may be cheaper in the long haul.  And, some people
and/or organizations may prefer to go that route without considering
the used route, so as not to have to take a gamble on how long the
hardware will last, or more importantly, how much it will cost them to
keep replacing it when it dies.

This argument ignores several factors, including: the cost of labor
for replacing the hardware, if your staff does it, vs. having a
warranty service technician do it for you; the cost of down-time when
the hardware fails; the cost of shipping the part (e.g. when
purchasing a used one, or when shipping one to be repaired or
replaced); whether or not a used card is or isn't more likely to fail
than a new card, on average.  The reasons for ignoring those factors
are that in many cases, they can be assumed to be the same in both
cases, and in many cases, they are too variable to consider directly.
For example, your vendor may pay for shipping the defective part back
to them.  Many vendors DON'T do that.  In the latter case, you could
assume the cost for shipping would be the same (per replacement), so
it would cancel out.  In the former case, you'd clearly need to factor
in the cost of shipping the used part, since there's no shipping on
the warranty part.  Ultimately, you'd need to do a more careful
analysis after having all the facts.


- -- 
Derek Martin               [EMAIL PROTECTED]    
- ---------------------------------------------
I prefer mail encrypted with PGP/GPG!
GnuPG Key ID: 0x81CFE75D
Retrieve my public key at http://pgp.mit.edu
Learn more about it at http://www.gnupg.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE8o2n6djdlQoHP510RAi2MAJ4ytSkPaY5bGr25pP6PgIB6H58H9wCfS8iz
lRTGp78lOffVHoCS+zaj2u8=
=zWnW
-----END PGP SIGNATURE-----

*****************************************************************
To unsubscribe from this list, send mail to [EMAIL PROTECTED]
with the text 'unsubscribe gnhlug' in the message body.
*****************************************************************

Reply via email to