[rescue] AT&T 3B2 revival

[Sandwich Maker]

" From: Alan Hightower <alan at alanlee.org>
"  
" 
" On 2015-08-18 18:45, adh at an.bradford.ma.us wrote: 
" 
" > remember, the 3b architecture was developed for the controller of the
" > 3ess, and telephone switchgear lasts a *looong* time; we designed for
" > 20-year 24x7 uptime. the 4ess, also using a 3b20d main controller,
" > was made from '76 to '99, and there are still some 100 in service,
" > with plans for a next-gen 4ess. the 5ess, still in production, has a
" > 3b21d 'administrative module'.
" > 
" > the 3b20d and 3b21d are, in telcom style, completely double-redundant;
" > you can walk up to one and pull any board at random without affecting
" > operation. they, and their relevant intellectual property, won't be
" > obsolete any time soon.
" 
" I thought most of the current 3B2 lineage (4E/5E) was done in emulation
" on Sun machines? I'm a bit surprised at the notion of a WE32200 in
" silicon running in 2015. 

that may be true of the 3b2 per se, but that was just an
implementation of the 3b20 arch in lsi [the '20 was ttl].  and the lsi
versions never had the telco-reliability-level of machines that can be
fixed while running.  for that, you need the full redundancy of the
'bigger' members.

suns don't have that reliability.  i think an emulation would be even
less reliable.

nb. the 3b21 is a '95-era machine; it may well be lsi instead of ttl.
even a gate array implementation would probably be a speed/power
boost.

ps. my first unix login was on a 3b20, running sVr0.  iirc it was
about half the speed of a vax 780.  my first formal unix admin was on
3b2s.
________________________________________________________________________
Andrew Hay                                  the genius nature
internet rambler                            is to see what all have seen
adh at an.bradford.ma.us                       and think what none thought