[rescue] oh please. (was: Re: OH YEA??? [was: Re: Ultra?])

[rescue at sunhelp.org]

> >   I don't know anyone who does "light" development work.
> That may be the case, but that doesn't mean that no one does it.

Alright, how 'bout ultra-light "development" work.  Time to
recompile BIND 9 from source on SS10 under Solaris 2.5.1 (one
of my compile hosts) with OpenSSL support and all the foo-forah
is > 1 hour.  Significantly greater, although I won't give a
specific number since I haven't timed it recently.  This is
(IIRC) a 2xSM41 system with 64MB RAM, although it *might* be
a 2xSM51 with 128MB RAM.  I "can't be arsed" to bother checking
right now.  Yes, we maintain a slightly-forked codebase
incorporating non-standard patches (PRNGd support, etc.) so
I can almost call this "development" since no one else could
build these binaries for us.

Meanwhile, a recompile from source under Solaris 7 on a U5/270
with 64MB RAM takes less than 15 minutes.

I'm sorry, but this is a no-brainer.  I'm not ditching the
SS10 any time soon...but I'm not running out to buy a pallet
load of them either.

And yes, both systems run under ~nil load most of the time,
and the U5 hasn't gone (down, or south) any more often than
the SS10.  Pretty much the limiting factor for uptimes in both
cases is site power, since neither is so critical as to require
a UPS (they're compile hosts...)

The limiting factor for basic usefulness of these two systems
is the tradeoff between raw speed and internal storage on the
one hand, with lower raw speed and external storage/tape drives
on the other.  Both have their places in our lab.

George, I respect most everything you say...but I think you're
all wet on this one and I guess I agree with the Rev. Chris H.
that an U5 has its place.

  --Rip