Re: [Diffserv-interest] Re: Diffserv-interest digest, Vol 1 #113 - 3 msgs
Brian E Carpenter <brc@zurich.ibm.com> Mon, 10 November 2003 19:52 UTC
Received: from optimus.ietf.org ([132.151.1.19]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA08778 for <diffserv-interest-archive@odin.ietf.org>; Mon, 10 Nov 2003 14:52:18 -0500 (EST)
Received: from localhost.localdomain ([127.0.0.1] helo=www1.ietf.org) by optimus.ietf.org with esmtp (Exim 4.20) id 1AJI4v-0004qW-7o for diffserv-interest-archive@odin.ietf.org; Mon, 10 Nov 2003 14:52:01 -0500
Received: (from exim@localhost) by www1.ietf.org (8.12.8/8.12.8/Submit) id hAAJq1m8018617 for diffserv-interest-archive@odin.ietf.org; Mon, 10 Nov 2003 14:52:01 -0500
Received: from localhost.localdomain ([127.0.0.1] helo=www1.ietf.org) by optimus.ietf.org with esmtp (Exim 4.20) id 1AJI4u-0004pl-Cy; Mon, 10 Nov 2003 14:52:00 -0500
Received: from odin.ietf.org ([132.151.1.176] helo=ietf.org) by optimus.ietf.org with esmtp (Exim 4.20) id 1AJI4C-0004ol-35 for diffserv-interest@optimus.ietf.org; Mon, 10 Nov 2003 14:51:16 -0500
Received: from ietf-mx (ietf-mx.ietf.org [132.151.6.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA08720 for <diffserv-interest@ietf.org>; Mon, 10 Nov 2003 14:51:01 -0500 (EST)
Received: from ietf-mx ([132.151.6.1]) by ietf-mx with esmtp (Exim 4.12) id 1AJI49-0003Ho-00 for diffserv-interest@ietf.org; Mon, 10 Nov 2003 14:51:13 -0500
Received: from imhotep.hursley.ibm.com ([195.212.14.170] helo=mail-gw1.hursley.ibm.com) by ietf-mx with esmtp (Exim 4.12) id 1AJI48-0003HJ-00 for diffserv-interest@ietf.org; Mon, 10 Nov 2003 14:51:12 -0500
Received: from localhost.localdomain ([127.0.0.1] helo=mail-gw1.hursley.ibm.com) by mail-gw1.hursley.ibm.com with esmtp (Exim 4.12) id 1AJI3b-0005KC-00; Mon, 10 Nov 2003 19:50:39 +0000
Received: from [9.20.136.27] (helo=sp15en17.hursley.ibm.com) by mail-gw1.hursley.ibm.com with esmtp (Exim 4.12) id 1AJI3b-0005K9-00; Mon, 10 Nov 2003 19:50:39 +0000
Received: from zurich.ibm.com ([9.145.244.160]) by sp15en17.hursley.ibm.com (AIX5.1/8.11.6p2/8.11.0) with ESMTP id hAAJobv49676; Mon, 10 Nov 2003 19:50:38 GMT
Message-ID: <3FAFEBE6.52164C4@zurich.ibm.com>
Date: Mon, 10 Nov 2003 20:49:58 +0100
From: Brian E Carpenter <brc@zurich.ibm.com>
Organization: IBM
X-Mailer: Mozilla 4.79 [en] (Windows NT 5.0; U)
X-Accept-Language: en,fr,de
MIME-Version: 1.0
To: "John H. Shuler" <johnshuler@mac.com>
CC: diffserv-interest@ietf.org
Subject: Re: [Diffserv-interest] Re: Diffserv-interest digest, Vol 1 #113 - 3 msgs
References: <BBD52302.14B11%johnshuler@mac.com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: diffserv-interest-admin@ietf.org
Errors-To: diffserv-interest-admin@ietf.org
X-BeenThere: diffserv-interest@ietf.org
X-Mailman-Version: 2.0.12
Precedence: bulk
List-Unsubscribe: <https://www1.ietf.org/mailman/listinfo/diffserv-interest>, <mailto:diffserv-interest-request@ietf.org?subject=unsubscribe>
List-Id: Differentiated services general discussion <diffserv-interest.ietf.org>
List-Post: <mailto:diffserv-interest@ietf.org>
List-Help: <mailto:diffserv-interest-request@ietf.org?subject=help>
List-Subscribe: <https://www1.ietf.org/mailman/listinfo/diffserv-interest>, <mailto:diffserv-interest-request@ietf.org?subject=subscribe>
Content-Transfer-Encoding: 7bit
Yes, WFQ is an example solution of course, but it's a relatively easy one to think about. I do agree we need a best practices document; we're only arguing about its richness in service classes, I think. Brian "John H. Shuler" wrote: > > Brian, > > I don't believe there is a mandate for SPs to use WFQ within the AF class, > so either WFQ or strict priority queuing is an option. As well, the nature > and degree of priority -- in fact, the entire policy equation -- seems to be > left open, except that there is an implication that AF4 is given some higher > quality/priority than AF1, whether in a weighted or strict sense. > > It is my contention, however, that there should be some kind of specific > guidelines provided for this class, especially, since it is in AF where a > lot of the subtlety -- and confusion, and cross-carrier screwups -- can > reside. This is where I believe the proposed draft can have its biggest > impact. That is, not only in delineating further the treatment of AF within > class, but also the rules in assigning traffic types to different classes. > > The whole idea is that, without some kind of standardization, multi-service > IP will not take off as it should, because there is no way for a customer to > know that the video (for example) he hands to his regional SP will make it > to the other end in a consistent and reliable way. > > j > > > From: diffserv-interest-request@ietf.org > > Reply-To: diffserv-interest@ietf.org > > Date: Mon, 10 Nov 2003 12:00:03 -0500 > > To: diffserv-interest@ietf.org > > Subject: Diffserv-interest digest, Vol 1 #113 - 3 msgs > > > > Send Diffserv-interest mailing list submissions to > > diffserv-interest@ietf.org > > > > To subscribe or unsubscribe via the World Wide Web, visit > > https://www1.ietf.org/mailman/listinfo/diffserv-interest > > or, via email, send a message with subject or body 'help' to > > diffserv-interest-request@ietf.org > > > > You can reach the person managing the list at > > diffserv-interest-admin@ietf.org > > > > When replying, please edit your Subject line so it is more specific > > than "Re: Contents of Diffserv-interest digest..." > > > > > > Today's Topics: > > > > 1. Re: QoS in diffserv network (Brian E Carpenter) > > 2. Re: QoS in diffserv network (Zoltan.Balint@alcatel.be) > > 3. Re: QoS in diffserv network (Brian E Carpenter) > > > > --__--__-- > > > > Message: 1 > > Date: Sun, 09 Nov 2003 19:26:48 +0100 > > From: Brian E Carpenter <brc@zurich.ibm.com> > > Organization: IBM > > To: Feng Y <feng6@uwindsor.ca> > > CC: diffserv-interest@ietf.org > > Subject: Re: [Diffserv-interest] QoS in diffserv network > > > > below... > > > > Feng Y wrote: > >> = > > > >>>> > >>>> It is up to the carrier to monitor congestion > >> periodically to prevent > >>>> lowest-priority traffic from being too congested. EF > >> should ALWAYS get > >>>> through on-time, and AF CIR should always get through, > >> if somewhat delayed > >>>> and jittery. The cases when this is NOT true should be > >> rare and bizarre > >>>> (earthquakes, major network outages, etc.) unless the > >> carrier is not doing > >>>> their job... Which is simply to make sure they engineer > >> total network > >>>> bandwidth above the level of peak CIR. > >>> > >>> Exactly. Each traffic class (i.e. each queue) needs to be > >> provisioned for > >>> the expected load with appropriate over- or > >> under-provisioning. The new > >>> thing is that you can over-provision for some traffic and > >> under-provision > >>> for other traffic. (New to IP, that is. It's hardly a new > >> concept.) > >>> > >> = > > > >> Can we always assume that a higher service class provides > >> the same or a better service than any lower service class > >> in AF? = > > > > > > No. There is no "higher" or "lower" built into AF. There > > is simply the option of having one or more mutually > > independent AF service classes, with an arbitrary number > > of four service classes being named AF1, AF2, AF3, AF4 > > if you choose to use the recommended code points. > > > > If you happen to give each of these classes equal weights > > in a WFQ scheme they are all offering the same level of service. > > > > If you happen to give AF2 a higher weight than AF3, you are > > choosing to offer more service to AF2 traffic. > > > > > >> Can the following scenario happen in Diffserv > >> network? > >> The gold service was allocated x percent of a link=92s > >> bandwidth and silver service was allocated x/2 percent of > >> the link=92s bandwidth, but the traffic intensity of gold > >> packets was 10 times higher than that of silver packets. > > > > That can certainly happen, if you mean that the offered > > load to the gold service is ten times greater than the > > offered load to the silver service. > > > >> = > > > >> Should the ISP assure that there are not so many gold > >> packets in service? > > > > It depends what the SLA for gold and silver say, doesn't > > it? There is no rule. It's the ISP's business decision. > > > > My assumption and what I would recommend to an ISP is: > > > > If the offered load on silver equals the bandwidth assigned > > to silver, then it should all be transmitted, unless the SLA says > > otherwise. And the excess traffic offered to the gold service = > > > > will be at high risk of being dropped, depending on the utilization > > of the remaining (100-x-x/2)% of the line. > > > > Brian > > > > > > --__--__-- > > > > Message: 2 > > Date: Mon, 10 Nov 2003 09:21:15 +0100 > > From: Zoltan.Balint@alcatel.be > > Organization: Alcatel, > > To: Brian E Carpenter <brc@zurich.ibm.com> > > Cc: Feng Y <feng6@uwindsor.ca>, diffserv-interest@ietf.org > > Subject: Re: [Diffserv-interest] QoS in diffserv network > > > > <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> > > <html> > > <head> > > <meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1"> > > <title></title> > > </head> > > <body text="#000000" bgcolor="#ffffff"> > > Brian, Feng,<br> > > <blockquote type="cite" cite="mid3FAE86E8.3AE456E7@zurich.ibm.com"> > > <blockquote type="cite"> > > <pre wrap="">Can we always assume that a higher service class provides > > the same or a better service than any lower service class > > in AF? > > </pre> > > </blockquote> > > <pre wrap=""><!----> > > No. There is no "higher" or "lower" built into AF. There > > is simply the option of having one or more mutually > > independent AF service classes, with an arbitrary number > > of four service classes being named AF1, AF2, AF3, AF4 > > if you choose to use the recommended code points. > > > > If you happen to give each of these classes equal weights > > in a WFQ scheme they are all offering the same level of service.</pre> > > </blockquote> > > That would be the case when the relative load in those classes was > > equal as well. But, since some<br> > > classes can be more loaded than others, the service seen in different > > classes does not <br> > > relate to the index of the AF class. Unless, of course, the scheduler > > was strict priority, which does<br> > > not conform to the requirements in RFC2597 (minimum bandwidth).<br> > > <br> > > <br> > > Regards,<br> > > Zoltan<br> > > <br> > > <snip><br> > > </body> > > </html> > > > > > > > > --__--__-- > > > > Message: 3 > > Date: Mon, 10 Nov 2003 16:55:17 +0100 > > From: Brian E Carpenter <brc@zurich.ibm.com> > > Organization: IBM > > To: Zoltan.Balint@alcatel.be > > CC: Feng Y <feng6@uwindsor.ca>, diffserv-interest@ietf.org > > Subject: Re: [Diffserv-interest] QoS in diffserv network > > > >> Zoltan.Balint@alcatel.be wrote: > >> > >> Brian, Feng, > >> > >>>> Can we always assume that a higher service class provides > >>>> the same or a better service than any lower service class > >>>> in AF? > >>>> > >>>> > >>> No. There is no "higher" or "lower" built into AF. There > >>> is simply the option of having one or more mutually > >>> independent AF service classes, with an arbitrary number > >>> of four service classes being named AF1, AF2, AF3, AF4 > >>> if you choose to use the recommended code points. > >>> > >>> If you happen to give each of these classes equal weights > >>> in a WFQ scheme they are all offering the same level of service. > >>> > >> That would be the case when the relative load in those classes was equal as > >> well. But, since some > >> classes can be more loaded than others, the service seen in different classes > >> does not > >> relate to the index of the AF class. > > > > Zoltan, > > > > There is no such thing as an index; the digit in AF3 is just part of the name. > > > > If two AF classes have equal WFQ weights but different offered loads, they > > will > > experience different loss and jitter rates, but the service *offered* is the > > same. > > The service *delivered* will vary, but that is a different matter. The two > > SLAs may > > be different, by allowing different levels of overbooking, but that is why the > > diffserv > > documents distinguish between the SLA and the SLS [service level > > specification]. > > > > I think we are in agreement, but I want to be precise. > > > >> Unless, of course, the scheduler was strict priority, which does > >> not conform to the requirements in RFC2597 (minimum bandwidth). > > > > True. > > > > Brian > > > > > > > > --__--__-- > > > > _______________________________________________ > > Diffserv-interest mailing list > > Diffserv-interest@ietf.org > > https://www1.ietf.org/mailman/listinfo/diffserv-interest > > > > > > End of Diffserv-interest Digest > > _______________________________________________ > Diffserv-interest mailing list > Diffserv-interest@ietf.org > https://www1.ietf.org/mailman/listinfo/diffserv-interest -- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Brian E Carpenter Distinguished Engineer, Internet Standards & Technology, IBM NEW ADDRESS <brc@zurich.ibm.com> PLEASE UPDATE ADDRESS BOOK _______________________________________________ Diffserv-interest mailing list Diffserv-interest@ietf.org https://www1.ietf.org/mailman/listinfo/diffserv-interest
- [Diffserv-interest] Re: Diffserv-interest digest,… John H. Shuler
- Re: [Diffserv-interest] Re: Diffserv-interest dig… Brian E Carpenter
- Re: [Diffserv-interest] Re: Diffserv-interest dig… John H. Shuler
- Re: [Diffserv-interest] Re: Diffserv-interest dig… Jing Shen
- Re: [Diffserv-interest] Re: Diffserv-interest dig… Brian E Carpenter
- [Diffserv-interest] Overbooking John H. Shuler