ard@boring.UUCP (04/22/85)
Hello, dear SUN-folks
Here (at last) the promised summary of my Global Inquiry about
the Sun Product Family. Sorry for the delay, Sun delivery times tend
to be quite long, you know! (alas, I had some holidays ...)
I have received 17 reactions of which 9 were serious discussing several
interesting topics. (4 reactions from Universities, 5 from Business)
I've ranked the issues with the the marks: ++ very good, + good,
~ moderate/normal, - unsatisfactory/bad, -- very bad.
*> 1 - general feelings/experiences with Sun equipment
++ Surprise (or not?): Everybody is generally very happy and satisfied
with their Sun equipment, except rusty, who is very negative having
had bad service and maintenance experiences (but he stands alone..)
Suns are believed to be good (enough) value for money.
*> 2 - hardware reliability, service, maintenance, delivery times, ...
+ This subject had the most responses. It turns out that reliability
of Sun hardware is high, esp. the FileServer (in the Ethernet
Sun network) was mentioned several times for its continuous
operation (at one place even 3/2 years without failures); there
seems to be a problem though with the disk cooling fans. Supplying
cold air solves it (indeed!).
Sun screens tend to blow up once every couple of months, one said.
But this problem seems to be (is?!) solved in the newer machines.
Service & maintenance generally OK.
Delivery times are LONG: 3 - 6 months. But they do deliver on time.
Adding device drivers can be (most often is) a nuisance: the
documentation is unsatisfactory (incomplete, sometimes wrong).
*> 3 - performance (eg. yes/no use of SunWindow: with how much memory)
+/~ Buy your Sun with 2Mb of memory. It hardly does anything
with 1 Mb. You need 2Mb to get reasonable performance using Sun-
Windows or any performance at all with SunTools.
Also, a sun with the 42Mb disk is felt to be pretty useless; 70Mb
is a minimum.
Suns are fairly fast machines, about as fast as a VAX-750
(both "unloaded"). Floating point speed is poor WITHOUT, and not
that fast (ie. slower than 750) WITH the extra FP-board.
SunWindows is reasonable, but needs adapting when used for user
applications.
There is little performance degradation when swapping/paging
over the Ethernet network to the FileServer (compared to a local
disk). However, if hardly any paging occurs, a Sun works best
(consider 3Mb, or even 4).
Note: only one 9600 baud terminal can ber attached to a Sun; others
must run at lower speeds.
*> 4 - yes/no problems concerning
+/~ No (big) problems...
> a. porting software to Sun UNIX (BSD4.2)
Porting software (from VAX) is painless, except the byteswap
problem, and (C) bitfields, unions and dereferencing a null pointer
(crashes on a 68000).
> b. Sun-1 Ethernet Interface on Sun-2 equipment
Ok.
> c. non-Sun boards/controllers etc.
Well possible.
*> 5 - Third Party software
+ Software used: Quintus Prolog, Unipress Emacs, Interleaf (technical
text processing tool), much USENET software.
*> 6 - any other (special) subject of interest.
. No specific responses.
Well, this is it.
Information I summarised above came from:
From: decvax!ihnp4!alberta!steve (Steve Sutphen)
From: decvax!ittvax!dcdwest!sdcsvax!sdcarl!rusty (Rusty Wright)
From: enea!erix!mike (Mike Williams)
From: enea!tut!hmj (Hannu-Matti Jarvinen)
From: seismo!PEREIRA@SRI-AI.ARPA
From: philabs!linus!security!denning!mbk (Mark B. Kadonoff)
From: decvax!sdcsvax!sdcsla!west (Larry West)
From: seismo!bdmrrr!shprentz (Joel Shprentz)
From: philabs!ihnp4!trwrb!trwspp!spp2!stassen
Thank you all who replied. I hope to have been of help to the net.ai and
net.works community; anyway it has made it clear to us to go for the Sun.
Bye to everyone, Ard. (mcvax!ard)
P.S. Ethernet deliberately isn't a Reg.Trademark... (is it?)
Other trademarks are: ... well, you know.hans@log-hb.UUCP (Hans Albertsson) (04/24/85)
I can add one little piece of info, on HW reliability. We had a disk
continually go bad on us during the first few months, in the manner
generally ascribed to heat problems. After replacing the thing completely,
all the problems seemed to go away permanently. We made NO change to
cooling arrangements. At all. Maybe a temporary overheat will permananently
damage a Fujitsu 169 MB?
This concerned a SUN 2/120 w 130 M disk space.
--
Hans Albertsson, USENET/uucp: {decvax,philabs}!mcvax!enea!log-hb!hans
Real World: TeleLOGIC AB, Box 1001, S-14901 Nynashamn,SWEDEN