[327] in resnet
Oversubscription
daemon@ATHENA.MIT.EDU (Gabriel David Custer)
Mon Nov 5 15:15:54 2001
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Disposition: inline
Message-ID: <sbe6a8ab.060@poplar.cedarville.edu>
Date: Mon, 5 Nov 2001 14:56:29 -0500
Reply-To: Resnet Forum <RESNET-L@listserv.nd.edu>
From: Gabriel David Custer <CUSTERG@CEDARVILLE.EDU>
To: RESNET-L@listserv.nd.edu
Currently we run a max of approximately 100 nodes (10/100mb switched)
per 100mb fiber uplink, with stacked 3Com 3300 edge switches and a
single Extreme Black Diamond switch at the core. Servers have a 100mb
copper connection to the Extreme switch. This seems to be working
okay in our situation. However, this figure has pretty much just
evolved and was probably more viable when we were running 10mb hubs at
the edge with the 100mb uplink. 100:1 sounds pretty high to me. We
have several locations that have or soon will be braking that 100 node
mark, so before we go purchase more 3300 fiber modules, I'm taking a
closer look to see if we should be measuring this some other way.
Judging by MRTG stats, it appears the utilization seldom goes above 4%
on any given uplink. Does anyone have some recommendations for number
of nodes per uplink, or what factors they take into consideration when
they increase their uplink bandwidth? What tools/software do you use
to measure the utilization?
Most servers run Netware 5.1 over IP, and most clients run Windows
2000.
Thanks for any suggestions!
Gabe Custer
Cedarville University
___________________________________________________
You are subscribed to the ResNet-L mailing list.
To subscribe, unsubscribe or search the archives,
go to http://LISTSERV.ND.EDU/archives/resnet-l.html
___________________________________________________
