Okay - I was gonna keep quiet on this - but -
==============================================================
Guy Smith <[log in to unmask]> said (among other things):
] Thus HP has created in me a mini monopoly. This may not have been their
] intent, but they would be foolish to change the scheme. I like your
] notion of a certification scheme for a price, but any price they put on
] that "service" would not match the income they receive from the drives.
{ I wonder - Novell comes to mind . . .
] Let's see . . . the newspaper tells me a local vendor is selling 1 gig
] SCSI drives for $700, and HP will sell me an MPE ready 1 gig SCSI drive
] for a mere $2,200, or about three times the market rate. I wonder if
] anyone like EMC is interested in sharing that $1,500 by reverse engineering
] whatever HP has planted in their drives and controllers?
] You touched on the one point concerning UNIX which may be the driver, and
] why nimrods like me don't mind paying the extra money for hardware. UNIX
] can, and at a whim will, dismember itself in a cybernetic form of auto-
] masochism. MPE doesn't.
{ You paint a vivid picture. :-)
] What will be interesting to watch is the rise (and fall?) of NT. With some
] fault tolerence built in, and with it inevitably driving hardware prices
] even lower, there may be some competition to UNIX for the small server
] system.
] And maybe more competition for MPE.
What Guy has failed to do is paint one item bright red: even though _I_ may
understand why the $1,500 should be spent, the decision maker doesn't. The
unfortunate thing is the decision maker decides on dollars. His sense of
relativity has been changed from Univac, NCR and Burroughs to Gateway et al.
=======================================================================
Phil Haseltine <[log in to unmask]> also said (among other things):
{ some very interesting and generally important figures concerning
{ disk drives which I generally agree with
] [log in to unmask]
] [log in to unmask]
] This response does not represent the official position of, or statement by,
] the Hewlett-Packard Company. The above data is provided for informational
] purposes only. It is supplied without warranty of any kind.
{ pet peeve - these 'disclaimers' make no rational sense to me
===============================================
Richard Gambrell <[log in to unmask]> said:
] Thanks to Phil Haseltine for helping to clarify what we get when
] we pay for an HP drive for our HP systems. It is *always* good
] to understand what you're buying.
Me too - really.
================================================================
Goetz Neumann <[log in to unmask]> added some comments including:
] Although not knowing it in detail, I think the big differentiator is
] the firmware of these drives' controllers. The mechanical stuff and
] phsyical bus specifications are identical, and the mostly identical
] design (or use of mass products parts) has lowered the per MB price
] for 'MPE discs' as well.
{ I was trying _not_ to get controllers involved
] The differences are features like the writing of special on-disc
] logs, that can be used for neat stuff like Predictive Support, and
] other diagnostics, that help HP not having to tell customers nasty
] words like INSTALL too often.
{ What's special here - the idea, channel, controller or drive? In
{ other words, where is feature implemented?
] Another very MPE specific criteria is that the disc must support
] a mode of writing to disc, where the host has complete control
] of the I/O, meaning it is able to wait for the completion of of
] a write operation to disc until it REALLY happened every byte in
] the sector, and not getting 'fooled' by the on-controller cache.
] Otherwise IMAGE logging to disc would be as worthless as the
] transaction manager's help to protect your system from data
] or file system corruptions.
{ They all do - or don't - depending on how you look at it. I was
{ told that from 79xx drives on, including cached drives, writes
{ were guaranteed. Autogenerator motors or some such? Was my data
{ _really_ at risk in those days?
] I also do not know whether the next PC store's SCSI drive provides
] a feature like AUTOSPARING, a self detection procedure of bad
] (or going bad) sectors.
{ I was under the impression this was part of SCSI standard, and
{ implemented at the device (drive) level.
] I would always go for the reliability.
{ Amen
============================================================
And finally, Flemming Rothmann <[log in to unmask]> said:
] I noticed that this subject has lead to questions in the direction of a
] multivendor hardware installation.
] In my opinion, one of three conditions must be met in order to practice
] this: 1) Certified hardware pieces, or 2) your company must be BIG so
] vendors have a commercial interest in keeping you happy, or 3) Be sure you
] know what you are doing and what the consequences might be.
{ Agreed, but IMHO single-vendor days are fading into the sunset
] If none of the first two conditions are met, you can be sure that if (or
] when??) anything goes wrong at your installation, the vendors will just
] point at each other saying "That's not my problem" - and YOU will be the
] looser. This is a fact of life - I've experienced it from our HP9000
] installation.
{ Haven't had a lot of trouble from HP. My standard ploy: use self tests,
{ figure best odds, tell vendor to fix it. If my call is wrong, I'll pay.
] Personally I like the idea of some kind of certification.
====================================================
Now for (what I sincerely hope is) my final ravings:
Definition: SCSI - Small Computer System Interface- A standard access
standard for computer peripherals defined by the ANSI committee in
1986 X3.131-1986. (I'd prefer better grammer, but that's what is in
my controller board manual).
I would think some progress has occured in the last 8-9 years - and
that they would call it SCSI-II.
Some advice (or least things to think about) for HP:
1) Don't waste your time telling me why I should spend $2,200 instead of
$1,500 - give me tools I can use to educate the decision maker instead.
2) Quit thinking of MPE and the 900 boxes as on an even footing with Unix.
MPE is an advanced operating system used to build reliable business
solutions, and can also run _properly written_ Unix application software.
3) If your hard drives have capabilities beyond SCSI say so - but be
prepared to defend your claims. Present them as a superior product
which includes SCSI as a subset. Even better, build controllers to
support the SCSI standard. Not everyone can afford a total solution
from HP all at once, and this may be an acceptable alternative (my
first Series II had ISS disk drives).
4) Make sure to work with other vendors to let your installed base take
advantage of advanced devices - whether there's an HP label on it or not.
You may recall that this all got started by a press release that said a
new drive system with apparently decent performance is on the market for
Unix boxes, but you might not be able to get it for MPE.
Finally, I happen to believe that MPE was, is and (hopefully) will continue
to be a superb solution to business computer systems. In my 18+ years
with MPE I have heard some really wonderful things users could do (and
did a few too) with the system. I'd hate to see it nose-dive because
HP didn't realize just how good it is.
And now, back to your regularly scheduled discussions. We apologize
for the lengthy interruption.
Regards -- Tony B. Shepherd -- [log in to unmask]
|
