> On Thursday, 04/13/2006 at 12:10 AST, "Edward M. Martin"
> <[EMAIL PROTECTED]> wrote:
> > This is somewhat of a vent.
> I can appreciate that the situation as it stands is untenable for
some.

While I'd be inclined to agree with Alan for larger customers, I think
Ed has identified a valid point, especially for people who currently
have relatively small systems, such as Flex boxen or MP2000/MP3000
systems. 

Currently, the smallest z800 or z890 configuration is approximately
$150K, without disk, tape, or any peripherals -- even if you go
ultra-cheap and get only SCSI disk and one ESCON tape drive. A workable
replacement hardware configuration for a very good Flex box *with* disk
and tape -- if it were permitted -- is in the $30K range, including
shipping and delivery. Given that the Hercules guys are getting 65-70
390 MIPS per Intel engine (better on Opteron), $10K will buy you a
pretty darn good quad Opteron emulated zArch-capable machine that can
easily get you into the 200 390 MIP range. 

The 800/890 solutions are a major investment in both hardware and
environmentals -- for example, in SNA's case, the transition between our
faithful MP3000 and the new Flex box we're acquiring is the difference
between 2 30A circuits and 25K BTUs of heat versus 1 15A circuit and 600
BTUs. We're getting *10 times* the amount of disk (at about $150/250G vs
$4000/9G disk on the MP3000) and the whole thing fits in 8RU. The
cheapest quote available for a crippled 1 CPU 800/890 was $95K, plus
another $60K for a used Shark with about a TB of disk. It consumes 200
sq ft of space, and requires 2 30A 110 circuits for the processor, and
at least one 60A circuit for the Shark. Makes it very hard to run the
desktops and the beer fridge when you have only a 250A service entering
the office...8-)

The direct result is that people who have a working small 390-based
solution now are pretty much priced out of the market. IBM currently
offers no cost-effective solution for these customers unless you are
somehow able to qualify for PWD (agreed, it's not *that* hard, but if
you take the no-commercial-work requirement seriously, you're basically
SOL). We're lucky to qualify for such a machine, but the folks who
aren't are pretty much screwed. 

> Naturally, this decision is even more painful if your CxO has not been
> working on TCO analyses and finding new opportunities to use the
> mainframe.  [Alas, it is true.  Some truly have not seen The Light.]

In the case of the smaller systems, I think the light has clearly been
determined to be an oncoming train. See above -- if I can't touch a
current VM or VSE system for less than $100K and I *still* have to buy
some disk and tape to support my users (and possibly upgrade the
building infrastructure as well as the hardware to support the HVAC and
power requirements just to stay current) when my current system is
costing less than $50K, then the8 direct result is migration of the
applications to other platforms and a 100% loss of revenue to IBM for
that customer. 

This isn't new news, although I'd grant IBM (specifically Alan and Reed
Mullen) a lot of credit for trying to make this better (at least for
VM). Now we're dealing with the margin items, and I'm afraid IBM doesn't
have a good answer to this problem at the moment. It's OK with me if IBM
limits the 64-bit Flex licenses to genuine IBM hardware -- that'd be a
fair compromise -- it just seems like a bit of unnecessary arrogance to
say "sorry, we can't help you" when it's clear that IBM is *choosing*
not to help. 

> I'm not sure I'd agree that IBM is "driving them away" but would
instead
> assert that IBM is doing its best not to hold anyone back.  As a
company,
> IBM (rightly) invests its resources (money, people, assets) into areas
> that will provide profit and growth and I can only hope that they are
> targeting their investments properly. 

See above. There's clearly a missing piece here. How to make that case
may be more of an issue, but the problem is really there. 

-- db

Reply via email to