May 2003
Emulation hurdles get examined at
symposium
PA-RISC expert says design may be more complex, time to
market longer
While a European software company is working on a
prototype of an HP 3000 hardware emulator for a possible prototype
before years end, another firm based in the US believes the
project could be complex enough to take much longer. Gavin Scott of
Allegro Consultants used part of his talk during the latest Interex
Solutions Symposium to caution customers not to expect too much too
soon from anyones emulation efforts.
But Scott also added that he believes the time to
market for any software that transforms Intel-based PCs into HP 3000s
wont be a crucial factor in homesteading. For the next several
years, he said, a majority of the installed base will be purchasing
an ample supply of 3000 hardware of recent vintage, rather than
migrating.
Between and 50 and 80 percent of the total 3000
customers out there are going to stay on MPE at least for three to
five years, Scott said. As vice president of a company
thats offering MPE technical support and products for the 3000,
as well as considering an entry in the emulator derby, Scott has a
viewpoint biased toward the markets status quo. But he said he
can see some movement toward migration from customers who can
afford its cost.
There are going to be a lot of people working
on migrations during that time, and obsoleting the 3000, but
its primarily the large customers, Scott said. The
medium and smaller customers say their short-term plan is Tell
me how I can not do anything, how the homesteading stuff can be
viable.
Citing good availability of 3000 systems on the used
market, and three more years of support from HP, Scott added that
At the moment I think the 3000s future looks better than
it has in the last couple of years since HPs [end of support]
announcement.
Complex IO
Scott said during an explanation of what an emulator
could do that the thorniest parts of fooling an Intel-based PC into
behaving like a PA-RISC system lay in mapping to peripherals.
Where it gets really hard is when you have to
simulate the whole IO system, all the peripheral cards, and even
potentially the peripherals themselves that the MPE operating system
is expecting to see, he said. Thats where most of
the time will be placed doing emulator development.
At the meeting, customers and developers began to ask
HP engineers about the possibility of the vendor creating pseudo or
pass-through drivers to aid in an emulators IO. HPs Mike
Paivinen did not affirm that HP would do this project, but the idea
was not dismissed immediately, either.
Collaboration to share risk
Scott said that hes been in communication with
some engineers at SRI, the European firm which is already selling
software that emulates VAX hardware on Intels Pentium systems.
OpenMPE board chairman Jon Backus suggested that one way to tackle
the project might involve several companies collaborating on a
project which the OpenMPE group would fund through membership dues.
Scott said that funding an emulator is the most serious challenge, in
his view, at the moment. But whatever design route SRI takes,
hes wishing the prospective competitor the best of luck.
If they succeed, then more power to them,
Scott said. An emulator would be a valuable addition to the HP
3000 customers range of options, even more so as time goes on.
So an SRI emulator is better than no emulator if the rest of us
cant figure out how to fund the development of the darn
thing.
Scott said the 3000 marketplace would be more
comfortable with a free emulator that we could all use, but at
the moment the intersection of the people who have the capabilities
to do an emulator doesnt include those who are independently
wealthy.
Another company, Strobe Data, is marketing an HP 1000
emulator and had expressed an interest in doing an HP 3000 product.
Scott noted after the meeting that Weve had some fruitful
discussions with Strobe, and if there were something to collaborate
on I think wed get along with them quite well. Any effort
among all three firms looks doubtful, he added.
Strobe and SRI are competitors in the DEC
emulation world, Scott said, so that makes a three-way
collaboration unlikely though its possible that we could
offer assistance to both parties if were not doing our own
emulator and not in a full-blown collaboration with Strobe Data.
Im open to any and all discussions, but my impression is that
an SRI/Allegro collaboration is not one of the most likely
scenarios.
Competing with used systems
Emulators give the 3000 hardware user a future with
fewer limits, but customers will have a more certain source of
systems that will compete with software emulation: the growing pool
of used HP 3000 servers.
Scott noted this might make the early marketing of an
emulator tougher, since our competition in 2004 would be people
selling used N-Class systems on eBay for $5,000. Todays
$175,000 system could depreciate, and its going to be hard to
sell a piece of software that will have poorer performance than any
N-Class 3000.
HP did confirm that an emulator would not have to be
developed by next year to maintain the HP license commitment for
emulator-based MPE. This [HP license] draft has no
expiration, said HPs Jeff Vance, who helped write the
language in the HP proposal for new MPE licenses.
Support from third parties for an emulator is an
essential problem which OpenMPE can turn toward today, according to
Vance. Licensing issues with popular software vendors remain to be
worked out.
OpenMPE may be stalled on an emulator, but that
doesnt mean OpenMPE as an organization has to be stalled,
Vance said. There are other issues like third-party software
models, where OpenMPE could champion licensing deals from third
parties, whose licenses like those from Speedware, Cognos,
Bradmark, or Adager might be tied to a 3000. Those other
licenses might not automatically carry over today.
Getting initial investment from some kinds of
customers in an emulator project may be difficult, according to new
OpenMPE board member Donna Garverick. At her employer,
theyre squabbling about monitors right now, so spending
even $500 on an emulator that may never appear might be a really
tough sell.
|