FWIW, you actually get marginally better performance with a hub than you do
with a switch if you have fewer than about 10 PC's.  With few hosts, the
switching actually slows things down.  We tested it back when I was working
as a DBA because we would often stick our application server and database
server on an isolated LAN with each other to handle the data query traffic,
and a hub was always faster when there were only a couple of systems on it.

My $0.02.

- Kevin

                                                                                                    
                   
                    "Jonathan Hale"                                                                 
                   
                    <maclaoch@earthlink.ne       To:     "Jared" <jared@trios.org>, 
<kclug@kclug.org>                  
                    t>                           cc:                                                
                   
                    Sent by:                     Subject:     Re: Cluster update                    
                   
                    owner-kclug@marauder.i                                                          
                   
                    lliana.net                                                                      
                   
                                                                                                    
                   
                                                                                                    
                   
                    12/03/2002 07:49 PM                                                             
                   
                                                                                                    
                   
                                                                                                    
                   

> Steinkuhler, the 48-port hub donated by David Nicol,

Really, for adequate performance, we need a switched environment...
My suggestion for what to do with the cluster: Come up with an
algorithm/matrix for determining how long it would take a thousand
monkeys pounding on typewriters to synchronously type "the entire works
of Shakespeare", write a proposal for a federal grant, test it on the
Linux cluster, and use the proceeds to benefit some of the other
worthwhile projects we have previously discussed.  (Don't laugh--I heard
that someone recently got a grant to study how mermaids would procreate
if they existed)...
You gonna be at the meeting tomorrow night?