|
|
| Author |
Message |
| 16 new of 65 responses total. |
popcorn
|
|
response 50 of 65:
| 
Apr 8 12:29 UTC 1995 |
Arnold, that would be Truly Cool. Thanks for thinking about it!
Re 47: I got hold of a new copy of the "test" program yesterday.
I've got a half-written party wrapper in my directory. The next
step, at this point, is to write party motd messages and post
them in co-op for discussion.
|
steve
|
|
response 51 of 65:
|
Apr 8 17:39 UTC 1995 |
I won't get my hopes up about it Arnold, but since we are bottomfeeders,
we won't not look anywhere for sources of equipment. ;-)
I suspect that the hardware would still be too expensive, but it
never hurts to find out.
|
janc
|
|
response 52 of 65:
|
Apr 8 18:08 UTC 1995 |
I don't really see why you insist on bottom feeding. I would think it would
not be hard to raise the money to shift onto something current (though
likely not a Sun). When Arbornet started the drive to replace the Altos
they weren't in any better shape then Grex is financially. Probably worse
since they had less users then than Grex has now, and a large user base is
a large donor base. I think this business of replacing one obsolete
computer with another obsolete computer may not be very well advised. You
invest a lot of work and money, and a few years later you have to do it all
over again.
My old grandfather always said he was too poor to buy cheap stuff. I think
Grex may be as well.
|
mdw
|
|
response 53 of 65:
|
Apr 9 00:33 UTC 1995 |
It would certainly be possible to buy something much newer. The problem
is, spare parts, expansion, & money. We currently have 128M in grex,
mainly by virtue of its being obselete. In a newer computer, that 128M
would be 5 or 6 times more expensive. We currently have duplicates of
virtually every piece of hardware that comprises grex. So, when a
memory card flakes out, if there were any reason to suspect the CPU
card, etc., we can quickly swap cards and get back up. Given the same
situation on newer hardware; we wouldn't be able to afford nearly so
much RAM, and if something broke, we could be looking at potentially 2-3
days of down-time, easily. The upgrade path for the used hardware we're
using is also attractive; we were able to reuse considerable parts of
the sun-2 in upgrading to the sun-3; and we'll be able to do so again in
upgrading to a sun-4. That means we're not making expensive gambles
here; but cheap experiments. New, but completely incompatible hardware
leaves us locked into that hardware, and it could leave us in the
position m-net was for many years, of being so completely committed to
that hardware that there just wasn't any other hardware upgrade path.
Actually, we don't have all that many members relative to the size of
the user population; we've made little effort to go out and recruit
donations, and as a result, almost all of our membership base consists
of long-time ann arbor users. A lot of our internet users have no idea
how the system is funded, and sometimes think it's funded either by
their school, or through their internet fee. I'd like to see us give
those users more of a chance to participate in helping this system.
|
janc
|
|
response 54 of 65:
|
Apr 9 00:38 UTC 1995 |
I agree with that. We don't want to get into really high profile fund
raising, but users should at least know that we need donations. We
don't need to pressure people to donate, but we should find a way to at
least ask.
|
popcorn
|
|
response 55 of 65:
|
Apr 9 08:59 UTC 1995 |
<valerie vehemently agrees with jan's #52, and wishes
more of the Grex staff agreed with that too>
|
steve
|
|
response 56 of 65:
|
Apr 10 03:26 UTC 1995 |
Well, if we wanted to, we could have a fund raiser, and get some
new equipment. The question is, do we want to? Right now we have
a new CPU which is going to make Grex 2.5 to 3 times faster than
before. Once we've made the leap to this platform, we won't *have*
to recompile our binaries again, for a long time, if ever.
Once we're on this little Sun-4/200 bottom-feeder special, we can
just change out the hardware and plop ourselves onto ever faster
hardware. When we feel the pinch of the Sun-4/200 lacking enough
horsepower to serve us, we can upgrade to a Sun-4/400, which is
again 3 times faster, although we would have to change our memory
boards.
These two SPARC cards are fully compatible with our current VME
backplane hardware. The options after that are the various SPARCStations
out there, like a SPARCStation 2. That would be about the speed of
the Sun-4/400, more or less. There are LOTS of other SPARCs we could
jump to at this point, *all of which blow the wimpy AT-class {4,5}86
boxes out of the water*.
So we can spend $5000 on a much faster system which won't be
worth $2000 in two years, or we can get incremental speed increases
with a SPARC system and move freely from system to system as our
needs arise. And going to the SPARC means we can be A LOT faster
in the future. Don't be mislead with "mips" ratings, or anything
like that. The higher end SPARCs are "server" technology, meanind
that there are hardware assists to speed up the processing of data.
There is also the mention of software. With SunOS 4.1.4 out now,
which took what is arguably the best UNIX out ever, and applies *all*
the patches that have been created for it in the last several years,
you have an OS that is one of the most thouroghly debugged systems
in the world.
Now, if we want to have a fundraiser and get newer hardware,
we can. But our needs are far better served with Sun equipment
than most AT-class systems.
|
adbarr
|
|
response 57 of 65:
|
Apr 10 04:00 UTC 1995 |
re: All above - can we put this in "Dog and Pony Show" mode. This
can be "sold" to people with the funds to make things happen. I
can see how the "incremental" approach will be very attractive
to some potential backers. Have never taken the time to
work the numbers but it would be a good idea to chart
prcessors/money/system needs and buy at the best price as we
go? I think.
|
steve
|
|
response 58 of 65:
|
Apr 10 04:36 UTC 1995 |
Well, by taking the incremental approach we'll be spending as
little as we can for a particular system.
|
mdw
|
|
response 59 of 65:
|
Apr 11 22:55 UTC 1995 |
We also get the equivalent of a full-time service contract. On new
equipment that's generally very pricey, and it usually still entails
more waiting than we have today with our "self-serve" packrat approach.
|
nephi
|
|
response 60 of 65:
|
Apr 12 10:47 UTC 1995 |
So, who has been playing with the xxx modem? What is the matter with
it, anyway?
|
popcorn
|
|
response 61 of 65:
|
Apr 12 12:09 UTC 1995 |
I'm not sure anybody has been making changes to it since it was installed.
Its problem is that it's configured wrong. I think the modem itself is
configured OK, but Grex isn't configured to talk to it. There's a staff
meeting scheduled for tonight; we'll be talking about the xxxxx modem at
the meeting.
|
steve
|
|
response 62 of 65:
|
Apr 12 12:33 UTC 1995 |
I'm going over to the dungeon today and either fix it or kill it.
|
popcorn
|
|
response 63 of 65:
|
Apr 12 14:22 UTC 1995 |
<valerie hugs STeve>
|
nephi
|
|
response 64 of 65:
|
Apr 12 14:33 UTC 1995 |
I think a lot of people will be hugging STeve. 8*)
|
davel
|
|
response 65 of 65:
|
Apr 13 01:27 UTC 1995 |
Thanks, STeve - *especially* if you succeeded.
|