[UFO Chicago] syncing multiple workstations to one master

W. Scott Lockwood III scott at guppylog.com
Thu Sep 8 17:03:08 PDT 2011


Also look at DRBL. http://drbl.sourceforge.net/

--
W. Scott Lockwood III
-----Original Message-----
From: ufo-bounces at ufo.chicago.il.us [mailto:ufo-bounces at ufo.chicago.il.us]
On Behalf Of Jesse Becker
Sent: Thursday, September 08, 2011 4:45 PM
To: Calvin Pryor
Cc: ufo at ufo.chicago.il.us
Subject: Re: [UFO Chicago] syncing multiple workstations to one master

You can go either way:  thin(ish) clients with LTSP, or thick clients that
have full OS installs.

In some ways, you can trade pain now for pain later, not to mention costs.
Having a single large server for ltsp may reduce your admin costs, but you
have to have good hardware.  Or you can go for thick cilents with small
"policy server" for your cfg mgmt program.

Without knowing more, I can't say which is right.

You could have the clients connect to a VM 'server', but that means an even
larer server, and more complicated management.  The advantage, however, is
that you can test new server software, but leave the 'production' system
alone in its own VM.

Two other points:

1) If you go the LTSP route, you can start with a small server, and see if
that model will work for you.  If performance is a problem, then get better
hardware, but use what you have until you can show there's a problem.

2) The thick client model will scale more easily than the LTSP model, since
all the 'server' is going to do is feed small policy files, and possibly
push larger files around the network occasionally.  Most of the "work" is on
the client side, so when/if you double the size of the lab, you probvably
won't need a huge new server.

--
Jesse Becker

--- On Wed, 9/7/11, Calvin Pryor <calvinpryor at gmail.com> wrote:

From: Calvin Pryor <calvinpryor at gmail.com>
Subject: Re: [UFO Chicago] syncing multiple workstations to one master
To: "Jesse Becker" <jesse_becker at yahoo.com>
Cc: ufo at ufo.chicago.il.us
Date: Wednesday, September 7, 2011, 9:56 PM


On Mon, Sep 5, 2011 at 3:46 PM, Jesse Becker <jesse_becker at yahoo.com> wrote:

Ah, You're using LTSP--I didn't know that using that before my last email.
 That actually makes things a bit simpler, so far as package distribution
goes.

We haven't actually built the lab yet.
First question is do we use the 12 (or so) boxes as clients ? Or install a
full blown OS on each box ? 

Whatever we decide to install, the existing g/PXE boot server will make
quick work of it. Clients would require a beefy server to connect to. Full
blown OSes would require lots of management, unless we use some of the tools
suggested.

Next question is, if we use the boxes as clients, to we connect to LTSP or
VMs ? Probably woulda helped if I started the question like this.......I'm
pleading the fifth...on what grounds I don't know , but I'll think of
something good. 


-----Inline Attachment Follows-----

_______________________________________________
UFO Chicago -- Users of Free Operating Systems
Free Software Rules -- Proprietary Drools!
http://ufo.chicago.il.us/cgi-bin/mailman/listinfo/ufo
_______________________________________________
UFO Chicago -- Users of Free Operating Systems
Free Software Rules -- Proprietary Drools!
http://ufo.chicago.il.us/cgi-bin/mailman/listinfo/ufo


More information about the ufo mailing list