Re: Doc[1] Charting Networks

Valdis Kletnieks <VALDIS@vtvm1.cc.vt.edu> Mon, 09 November 1992 21:13 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa23342; 9 Nov 92 16:13 EST
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa23338; 9 Nov 92 16:13 EST
Received: from haig.cs.ucl.ac.uk by CNRI.Reston.VA.US id aa08775; 9 Nov 92 16:14 EST
Received: from bells.cs.ucl.ac.uk by haig.cs.ucl.ac.uk with local SMTP id <g.07372-0@haig.cs.ucl.ac.uk>; Mon, 9 Nov 1992 20:06:43 +0000
Received: from vtvm1.cc.vt.edu by bells.cs.ucl.ac.uk with Internet SMTP id <g.14987-0@bells.cs.ucl.ac.uk>; Mon, 9 Nov 1992 20:06:36 +0000
Received: from vtvm1.cc.vt.edu by VTVM1.CC.VT.EDU (IBM VM SMTP V2R2) with BSMTP id 6282; Mon, 09 Nov 92 15:05:17 EST
Received: from vtvm1.cc.vt.edu (VALDIS) by vtvm1.cc.vt.edu (Mailer R2.08 R208002) with BSMTP id 2663; Mon, 09 Nov 92 15:05:17 EST
Date: Mon, 09 Nov 92 14:51:29 EST
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Valdis Kletnieks <VALDIS@vtvm1.cc.vt.edu>
Organization: Virginia Polytechnic Institute
Subject: Re: Doc[1] Charting Networks
To: Glenn Mansfield <glenn@aic.co.jp>, spp@aic.co.jp
cc: ic-tech@isode.com, osi-ds@cs.ucl.ac.uk, thomas@aic.co.jp
In-Reply-To: <9211090220.AA07142@aic-wide.aic.co.jp>
Message-Id: <921109.145129.EST.VALDIS@vtvm1.cc.vt.edu>

On Mon, 9 Nov 92 11:20:08 JST you said:
> 5.2.3 Line
 ...
> line OBJECT CLASS
 ...
>   charge         :: numericStringSyntax,
>        /* to be paid by user for transmission
>             of data (per packet/byte/time...)  */

It's  crucial that  the units  be  carried along  here. There  is a  big
difference  between $2000/month  and $2000/packet.  Of course,  here you
can make a guess that $2000US is a  "per month" charge - but what if the
value is $2.50 - is this a "weekly" or a "per megabyte" charge?

Also, a numeric value alone isnt much  use. If the object described is a
T-1 link running  from an ENSS in Ithaca, NY  to Montpelier, France, are
the  units  US dollars  or  francs?  And  what effect  do  international
currency fluctuations have?

>   cost      :: numericStringSyntax,
>        /* to be paid to service provider for use of line
>             (per month, traffic...)  */
Same as above..

>   traffic        :: numericStringSyntax
>        /* (average) use in percent of nominal bandwidth */
>         }

What timeframe  is this "average" taken  over? And how do  you deal with
objects  with widely  varying loads  over time/space?  For instance,  we
have  a  large (about  800  node)  Ethernet-based network  that  employs
bridges  for traffic  filtering. Some  segments (for  instance, over  in
McBryde Hall,  where both Math and  CS are located) run  quite congested
during  the day,  but  are essentially  idle at  night,  except for  the
occasional NTP packet.  Other segments sit at a pretty  constant 15% the
entire day.

Also, remember that  for many media types, nominal  bandwidth has little
relationship to actual  capacity. Note that both  the Ethernet collision
detection/avoidance and  most routers perform somewhat  differently when
presented with a few 8K NFS  packets fragmented into 1500 byte segments,
an  several  zillion  telnet  packets  with  one  character  of  data  a
packet....


                                  Valdis Kletnieks
                                  Computer Systems Engineer
                                  Virginia Polytechnic Institute