Re: about QoS

Kevin DeLange <kdelange@lighthousecomm.com> Fri, 09 October 1998 13:32 UTC

Received: from ns.newbridge.com (ns.newbridge.com [192.75.23.67]) by ietf.org (8.8.5/8.8.7a) with ESMTP id JAA26883 for <qosr-archive@odin.ietf.org>; Fri, 9 Oct 1998 09:32:18 -0400 (EDT)
Received: (from smtpd@localhost) by ns.newbridge.com (8.8.8/8.6.12) id JAA01237 for qosr-archive@odin.ietf.org; Fri, 9 Oct 1998 09:32:13 -0400 (EDT)
Received: from kanata-gw1.newbridge.com(192.75.23.72), claiming to be "kanata-gw1.ca.newbridge.com" via SMTP by ns.newbridge.com, id smtpdBAAa00896; Fri Oct 9 09:31:42 1998
Received: from distmaster.ca.newbridge.com by kanata-mh1.ca.newbridge.com; Fri, 9 Oct 1998 09:14:12 -0400
Received: by distmaster.ca.newbridge.com (SMI-8.6/SMI-SVR4) id IAA25609; Fri, 9 Oct 1998 08:56:00 -0400
Message-ID: <361E0620.E2AE5733@lighthousecomm.com>
Date: Fri, 09 Oct 1998 08:48:32 -0400
From: Kevin DeLange <kdelange@lighthousecomm.com>
X-Mailer: Mozilla 4.05 [en] (Win95; I)
To: rick king <rickyy@bbs.huizhou.gd.cn>
CC: mpls <mpls@netlab.indiana.edu>, qosr@newbridge.com
Subject: Re: about QoS
References: <000801bdf2d1$f3c67c40$060a060a@dream.ritt.org.cn>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-qosr@newbridge.com
Precedence: bulk
X-Info: [Un]Subscribe to qosr-request@newbridge.com

Rick,

If the only reason you want to implement QoS is to insure that there are only
two classes of traffic moving over the Internet, then your proposal makes
sense.  However, there are numerous reasons why you need a more robust mechanism
to segment traffic.  Critical applications, and their underlying protocols, do
not always need the same traffic priorities.  It's not on/off, it's more like
how much do you want to turn it on.  More importantly, you have to consider the
difference between constant bit rate versus variable bit rate applications.  If
you want to deliver voice over a packet switched network, you need some
guarantees that the voice traffic will be provided a steady stream of priority
traffic.  At the same time you may need to guarantee that mission critical data
applications be given priority over other less sensitive traffic.  And if
multicast ever becomes a reality over the Internet, it will become even more
critical to offer a robust mechanism to allow for more effective bandwidth
utilization.

Kevin



rick king wrote:

> As I have know about MPLS,RSVP,Diffserv,they are all supposed to offer QoS
> over Internet.But all difficult to implement. I wonder why not use only one
> bit in IP header to indicate this packet's priority? Let's call this bit as
> P bit. If it's set to 1,then the packet has high priority.otherwise the
> packet is treat as usual.The router will keep the delay of high priority
> packet within a range.When congestion occur ,the router will discard the
> packet with P=0.
>
> If you set P to 1,then you will pay for it.Every host and router can set the
> P bit.if the router decide that the network can't support too much packet
> with P=1,it can override the P bit ,and you will not be charged.
>
> I think this method is easy to put in practice.any suggestion to it?