*BSD News Article 82443


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!howland.erols.net!EU.net!news2.EUnet.fr!newsbr.eunet.fr!usenet
From: Frederic.Marand@osinet.fr (Frederic MARAND)
Newsgroups: comp.unix.sco.misc,comp.unix.misc,comp.unix.questions,comp.unix.bsd.freebsd.misc,comp.unix.bsd.netbsd.misc
Subject: Re: which Unix to choose?
Date: Wed, 06 Nov 1996 21:02:52 GMT
Organization: Groupe SEDI / Agorus SA / OSI SARL
Lines: 122
Message-ID: <55r1u0$1mv@newsbr.eunet.fr>
References: <327F5727.179F@citynet.net> <327FC59A.6E48@gcsnet.com> <55phr2$5ln@newsbr.eunet.fr> <3280DA06.383@gcsnet.com>
NNTP-Posting-Host: 193.107.196.155
X-Newsreader: Forte Free Agent 1.0.82
Xref: euryale.cc.adfa.oz.au comp.unix.sco.misc:28689 comp.unix.misc:26119 comp.unix.questions:90666 comp.unix.bsd.freebsd.misc:30736 comp.unix.bsd.netbsd.misc:4825

"Ismaeel Abdur-Rasheed, et. al." <ismaeel@gcsnet.com> wrote:

>I have found Sun hardware to be remarkably reliable.  I have been
Notice I said nothing about Sun boxes, only Solaris.

>delighted with the technical competence of their hotline staff.  And,
>apart from the problem with ypserv, have not had any problems with
>Solaris since version 2.3.
...and to be more precise Solaris <= 2.3. This seems to match with
your observations.

>I consider the fact that Solaris releases come with accompanying patch
>CD's, and monthly patch CD additions to be a excellent form of support!
Basically, my point on this issue is an OS should not require patches,
because it should work Not work when fixed monthly with technical
support from the machine's manufacturer paid for dearly. Just work,
period. And that is exactly what - in our experience - AIX dellivers
and Solaris (<= 2.3) fails to deliver.

>In fact, the easiest thing to do, is when a patch CD is received,
>execute patchinstall, and simple select "suggested" patches, and off you
>go.  What could be easier.
This is simply unthinkable in a heavy production setting. You can't
just patch a live OS, reboot a machine, and so on, when tens of cash
registers are linked to it and customers lining up to pay with their
food caddies full. In this business, even one minute donwtime means
serious money.

>IBM, on the other hand, turns it's nose up at the entire unix industry,
>who after great effort finally consolidated their versioning around AT&T
>SVr4.  
This is (partly) true, and we may regret this attitude. Basically,
this SVR4 consolidation does not seem quite admitted when you consider
all the free Unix systems, which don't have normal SVR4 behaviour,
even down to a full fledged Korn shell.

>AIX, however, uses it's own proprietary methods of system
>administration (smitty).  
Smitty is not a proprietary method for administrating AIX. It is a
user interface for system management, much like SCO's sysadmsh,
Uniware graphical utilities, or HP's utility. There is indeed a
proprietary method underneath, based on a lot of ODM databases, but it
is certainly not smitty. This lower layer is certainly not standard,
but it IS interesting. It is consistent, table driven, most management
tasks can be performed using an identical management interface. But
the best part about it is, on top of this proprietary technology,
specific methods, they managed to build a lot of compatibility (the
real failure at compatibility is in the print spooler backend part). I
think this is a worthy enterprise, much more interesting than just
buying technology from Unix Labs (no, Univel ; no, Novell ; no SCO/HP
; no, who else ?) and reselling it. Don't you think so ?

>SO when you move from one unix box to another
>- if they are SVr4 you have a good chance of being able to administrate
>the system using the command line syntax; but if you move to a AIX
>machine - your sunk!
Most system management methods taken from SysV or BSD work fine,
whereas on other systems with a proprietary management interface,
messing with "standard" files can deadlock your machine, or lead to
uncomplete configuration.

>IBM is like Microsoft in that way.  Insisting on doing things with their
>own flavor and to hell with the rest of the industry.  Well, ever since
>Compaq build the first x86t clone, "we've had all we can stand and we're
>not going to take it anymore!"
That's what I used to think. Then I had to manage all these 6000s, and
found out it took me less time for these hundreds of machines that
just to keep my own Win3.x (then NT3.1, then 95) running without
GPFing several times a day.

By the way, what do you think Sun does ? It bought full rights to
Unix, and just keeps on devising its own nonstandard extensions. 

And DEC ? They hopped on the wrong bandwagon with OSF/1. Interesting
beast, though: this is the only Unix I saw with my own eyes run on the
same day on a Sparc 1, a RS/6000, a DECstation, an Apollo DN, and
several forgotten machines. Best processor independence,
multithreading since its inception, and noone got caught on it.

And HP ? They just bought Unix itself, like Novell. Let's hope it will
do the system more good than the Novell adventure...

>What happened with Microchannel?  The same story.  IBM says "jump", and
>anyone with half a brain says "f**k you!".
Microchannel had several problems:
- too low power capabilities for cards
- too demanding form factor for the time when it was released
- and, worst of all, unsane licensing terms.

Given its technical superiority over the PC buses of that time, it
could have worked if the license had been 5 to 10 times cheaper.
Remember the 16-bit ISA bus performance, the 32-bit MCA one, compare
that with the limitations of the early 32-bits buses on PCs, like
those produced by AST, Compaq, and others, and even EISA. It took PCI
to finally overcome the technological leap MCA could have meant when
it was so poorly released by Big Blue, and just try to remember how
many chipset versions it took Intel and others to finally get PCI
stable, when MCA systems ran properly from the first day onward.

You might also mention the 32-bit SIMM, while you're at it : PS/2s
could use all combinations of any number of single 32-bit SIMMs of
(almost) any capacity, when clones were always limited to banks of
same size, except at the end of the 486 epoch with 32-bit SIMMs, in
1994/95. Those machines could manage it 7 years earlier ! 

Well, you had to pay for it, and maybe it was really NOT worth it. But
this is mostly poor policy from IBM. The underlying technology is
often superb.

You can also consider disks, if you care to admit most head techniques
today revolve around IBM's MR head technology. And I suppose the list
could go on forever, or almost.

>(flame, flame)
>>>>>>>>>>>> how many megabytes of it will come, do you think ?

---------------------------------------------------------------------------------------------------------------
An AIX fan. Sun reseller, too... nobody is perfect.
---------------------------------------------------------------------------------------------------------------