>I guess by the latter, since artificial feature blocking is one of many
> evils of the so-called "market economy" which this oppresive government
> imposes.

Propagandist crock.  If it's in the box it's available.  All you have 
to do is drive the hardware.  The so called "bit" is a simple register 
that can be read or ignored and it happens that VMS reads it and uses it
as does ultrix...they don't have to though.  I'd likely bet to netbsd 
it's a don't care.

<>    True, but they can't be THAT bad. KA42-Bs are even better.
<
<You're not gonna be competing with Pentiums speed-wise on a KA410.

Vaxen (at least running VMS) slow down gracefully.  Even KA42 is a 
somewhat limited system user capacity wise.  I tried to not use them 
as hosts with more than say 30 or so users (especially if they were 
using xterminals).  The larger BIbus vaxen do better but it's more a 
factor of storage IO performance and Eithernet limitations.  

For a given CPU the 3100s bounded boxes(m10/20/30/38/76) are the lower 
performance of the lot with the Q-bus and BIbus versions of the same CPU 
offering better performance from faster/smarter disk interfaces.  While 
impressive the 3100m76 in real life applications rarely outperformed a 
3500.  For internet services if the pipe from the net is narrow enough 
(say a single eitherent) the rate at which requests come are limited and 
the smaler vaxen will perform ok.  Of course if you find a 3100m80 or 90 
it will do better but a low end 4000 or 6000 series box will be better 
still and likely occupy less space and consume less power. Again my first 
hand experience is with VMS systems with eithernet as the primary user 
interconnect.  I'd suspect that Unix would make little difference save for 
a possible different division of task assignment.

The VS2ks however are out performed by microvax-IIs in Q-bus without 
question.  The disks in the Q-bus space are hosted by either faster or 
more sophisticated controllers designed to relieve the CPU. Also using 
multiple controllers to split loading is effective in qbus boxen.  A 
MV-II with two RD54 disks one per RQDX3 will perform better than a single 
RQDX3 for example.  Same for SCSI controllers(unless they do local 
caching).  Some of the faster DEC controllers like massbus are fast 
enough that the disks are no longer the bottleneck and the CPU or other 
IO are the limiting factors.  Experience has shown me that VS2ks make 
excellent single user or limited multi user systems (fine for MOPbooting 
servers for example) but a MV-II or CVAX in Q-bus offers much better 
capacity never minding expandability.

If I were building a net around Vaxen in BA123s(or sboxen), 3100s and 
VS2ks their useage would be:

VS2k local net terminal services, boot loading servers
Q-bus MV-IIs/Cvax routers, bridging, firewalls, disk farms mailservers,   
                nameservers, FTP/fileserver
3100s(depending on model) Webserver, limited disk farms, mailservers,
			  FTP/fileserver


Allison