Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 08 Oct 1999 15:28:37 -0600
From:      Brett Glass <brett@lariat.org>
To:        Joseph Scott <joseph.scott@owp.csus.edu>
Cc:        chat@FreeBSD.ORG
Subject:   Re: Targeting the server: Not such a good idea?
Message-ID:  <4.2.0.58.19991008143949.043f2a30@localhost>
In-Reply-To: <37FE3827.A45735D8@owp.csus.edu>
References:  <4.2.0.58.19991008083634.044de740@localhost>

next in thread | previous in thread | raw e-mail | index | archive | help
At 06:29 PM 10/8/99 +0000, Joseph Scott wrote:

>         Before I comment on these comments, I believe everyone involved in
>FreeBSD would love to see FreeBSD be much more comparable to Windows (
>support, apps, drivers, etc ) in the desktop world.  However being that
>resources are finite then tough decisions must be made.

It is true that resources are finite. But as FreeBSD grows in applicability,
the pool of resources available for its development will also expand.

>   FreeBSD comes
>from a strong server background and chose ( I believe correctly ) to
>play to that strength.  That being said it has not totally forgot the
>desktop market ( that's just not the main focus ).  I've been running
>FreeBSD on my desktop at work ( as well as servers ) and I'm very happy.

I run it on several desktops and am happy with it too. But if it is to
reach the mainstream, it mustn't CLAIM that it's designed only for 
servers! If it does, it is literally telling the world it has
limitations it does not really have. And, seeing that Linux does not
claim such limitations, users will choose Linux.

I'm reminded of the quote from Richard Bach's "Jonathan Livingston
Seagull" -- "Argue for your limitations, and you get to keep them."

>To compare Novell's problems with FreeBSD isn't really 1:1 I believe. 

No analogy is perfect. NetWare, in fact, was in worse shape because it
really could NOT be changed to address the desktop. The irony is that
FreeBSD CAN, but is telling the world that it cannot or will not.

>As I see it Novell blew it because it took them way too long to switch
>to IP.

They actually had NFS before the Internet became privatized; they could
handle clients which used IP. And routing of IPX wasn't a problem;
Cisco supported it VERY early on. Print service was available
to IP clients too. So, the two main things that Netware did --
file and print services -- were covered. Web service came as an
NLM as soon as the Web began to grow. Netscape's Commerce Server for
Netware is quite good, I'm told.

>   Inspite of that fact Novell still has a pretty healthy install
>base, and most Novell admin.'s would probably tell you they would like
>to continue to use it, but Novell didn't catch up to the Internet age
>fast enough.  I would think that if you were to launch a large survey to
>find out from (ex) Novell admins why they switched none of them would
>mention anything that dealt with desktop issues.

That's because NetWare could NOT work on the desktop. But if they switched
their servers to NT, customers could run the same thing on the server as on
the clients. That was Microsoft's promise, and it was just what the
admins wanted to hear. Microsoft won big.

> > "The trouble is that no one wants to have separate training, separate
> > software, separate configuration, or separate experts for the server.
>
>         An IS person said this?

Not just one IS person -- LOTS of them. Stewart Alsop said it at the Senate's
hearings on Microsoft. Gates says it regularly in his pitches to CIOs and
IS people, and they like what they hear.

>   Wow, where did this person go to school, I'll
>be sure to avoid it.  Seriously, having an IS department is a fact of
>business.  

Of course it is! Using the same OS on the client and the server does not
and should not eliminate the IS department. But it does make its work MUCH
easier. 

> > Companies are tired of paying a CNE 'guru' big bucks to fix NetWare and
> > then keeping a whole separate staff around to support DOS and Windows. And
>
>         Nothing has changed here.  If they are going to have a server OS they
>will need someone who knows who to keep it going, that's true of NT,
>VMS, unix, Novell, any NOS will need someone who knows it.  

Agreed. But it is easier to find that expertise, and keep it around in house,
when there are fewer OSes to support. The depth of expertise is also likely
to be greater. And problems which might arise because the "server expert"
doesn't understand the client -- or vice versa -- go away! It's really
tremendously appealing. I see this desire constantly at the sites where
I consult.

> > they want their workstations to act like servers: fast, efficient, and rock
> > solid. They don't want to see these traits limited to the server! The
>
>         I would disagree here also.  If this were the case OS/2 would have 100%
>of the desktop market.

OS/2 failed for many reasons. It emulated Windows, which killed native 
application development; it wasn't multiuser, which limited its security 
paradigm and required access controls to be "glued on" after the fact; the
event handling was single-threaded (primarily due to the ignorance of
Microsoft developers in its early days, but the problem was never corrected);
it was poorly marketed; the press and system vendors were bullied by Microsoft 
into shunning it; and there were factions within IBM itself which sought to 
kill it in favor of AIX, Windows, or NT. 

>   Just because something it more "advanced"
>doesn't mean that it will win ( see my point about OS/2 ? ). 

OS/2 is advanced in some ways (particularly the Workplace Shell GUI), but
has always been behind the curve in others.

>Determining if a product will be successful has other issues besides
>being the most advanced product.  Microsoft has learned this point well,
>they've done their homework and hire people who know how to do
>"business", not so much computers.

Which brings us back around to what CIOs and IS managers want: one OS
on the desktop and the server.

>         On this point you do have to give Microsoft some credit, because of
>them and Apple the computer has become more assessable to more and more
>people.  The unix world is racing towards the center to gain ease of
>use, the windows world is racing towards the center to gain stability (
>not unlike politicians racing to the center as the election gets closer
>).  It remains to be seen who will win.  Unix has been making bigger
>leaps than windows I think, but there really is still quite a bit of
>distance to go.  NT on the other side has a lot of baggage that is
>making it hard to move towards the center.

I'm not sure what you mean by "the center" in the analogy above. Could
you elaborate?

> > operating system that everybody wants will run on everything, maybe with a
>
>         Now this I agree with.  I would love to have an OS that will run
>everything for me.

Me too! This is the big brass ring that NT and Linux are going for. FreeBSD
should be in the running for it as well, since it has advantages over
both.

> > few tweaks for what it's doing, and will be reliable, fast, and secure
> > everywhere. And if you know how to fix the desktop machine, you will be
> > able to work on the server too. NT would have taken over the world by now
> > if it weren't so insecure. It took over my organization just because
>
>         The type of people who only deal with "desktop" systems aren't in jobs
>to work on "servers", they are there to actually work on company issues,
>not internal items.  It's simply not possible to have one generic type
>of person and expect them to fill every possible position.  

People will always specialize. But the less disjoint we make the skill 
sets required to manage the client and the server, the more we benefit.
Less retraining, less confusion, less redundant staffing, better use
of people's time -- even better career paths, as people move more easily
from the help desk to serious administration.

>I would hate
>to see what would happen if my work told I now also had to do all of the
>accounting, I'm afraid my two courses in accounting in college just
>wouldn't be enough for me to do that job well.  The same is true of
>expecting every desktop user to be able to take managing a "server".

Whoa; let's be realistic here. We wouldn't expect every USER to manage
a server; however, it's a big plus if the knowledge used to manage a desktop
is directly applicable to the server and vice versa. 

Let's consider an example. To manage the IP stack on a Windows client, you
have to know the commands "winipcfg" and "route print" -- plus all their
idiosyncrasies and command line switches. You also need to know your
way around the Network Control Panel applet -- for example, that you should 
ignore Microsoft's annoying messages which state that the network is "not 
complete" if you turn off NetBIOS. (Leaving NetBIOS on is actually a big 
security risk, so Microsoft's messages goad administrators into making
clients insecure. You have to KNOW that the system will work better
if you ignore the message. MCSEs, as a rule, do not.)

On a UNIX server, you might use "ifconfig -a" and "netstat -r" (which also 
have their own peculiarities) to manage IP configuration and routing. There are 
also bunches of daemons to know about, including inetd (which may or may not 
be serving as a front end for the others).

Wouldn't it be nice to teach your network admins one set of commands,
rather than requiring them to speak (and be trained in) multiple dialects?
I must admit that I get some consulting hours due to this Tower of Babel, 
but I'd rather be doing other things.

>         Now about this statement about NT taking over and security.  This makes
>me seriously question your friend.  All operating systems having
>security issues.  Unix has been around for along time and has spent a
>lot of time cleaning things up.  NT doesn't have as much history, but it
>to is trying to clean things up.  Aside from the debate as to the
>ability of MS to clean up NT's security issues, the fact is that both NT
>and unix have had issues.  Some argue that NT's default security is
>crummy ( and I would agree with them ), but on the other side, how many
>Linux distro's don't have shadow passwords enabled by default? 

This is one of the reason I favor the BSDs. I very much like the way
that OpenBSD is super-secure by default, for example.

>  I believe that part of the reason that NT has not taken over in the NOS
>market is because of security issues, but to blame it entirely on that
>shows a lack of understanding, IMHO.

When I'm out in the field, I hear what my clients have to say. Security is
far and away the BIGGEST reason NT has not taken over. It dominates every
other issue.

> > Microsoft *promised* that it would run on both the server and the desktop.
> > The OS graveyard is littered with the bones of OSes that wanted to be
> > server-only. If FreeBSD is going to make itself look like it's only for
> > servers, the same thing will happen to it."
>
>         This may or may not be true, but mice show up when I leave food out,
>but I can't create mice by leaving food out.  

You don't need to create the mice; there are PLENTY out there already! And
if you feed them, they'll create more for you. ;-)

>I don't believe that the
>arguments of your friend point to anything that could be considered
>constructive.  Having said that I believe there are some very valid
>arguments for increased desktop support, it's just that your friend
>didn't come up with any of them.

What arguments for desktop support do YOU believe are valid?


>         In all fairness I believe that the reason your friend see's things this
>way is because of his experiences and issues.  From his point of view
>the items he brings up may be very true, however my experiences have
>obviously been very different than his.

As always, different strokes for different folks. But I think that all of
the evidence points back to the same thing: Forsaking the desktop is not
a good policy.

--Brett



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-chat" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4.2.0.58.19991008143949.043f2a30>