<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7652.24">
<TITLE>RE: [cisco-nas] Support for "percent" directive in per-session QoS</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>Many thanks, I shall give this a try, as with many new features these days documentation is sparse on useful stuff such as this :)<BR>
<BR>
<BR>
------------------------------------------------<BR>
David Freedman<BR>
Group Network Engineering<BR>
Claranet Limited<BR>
<A HREF="http://www.clara.net">http://www.clara.net</A><BR>
<BR>
<BR>
<BR>
-----Original Message-----<BR>
From: Ben Steele [<A HREF="mailto:ben@internode.com.au">mailto:ben@internode.com.au</A>]<BR>
Sent: Thu 1/10/2008 00:10<BR>
To: David Freedman<BR>
Cc: cisco-nas@puck.nether.net; cisco-bba@puck.nether.net<BR>
Subject: Re: [cisco-nas] Support for "percent" directive in per-session QoS<BR>
<BR>
They sure are, they will just reference the bandwidth you specify in<BR>
your radius av-pair, so if you set priority percent 10 to a class you<BR>
have put 435Kb in radius for it will give up to 43Kb of LLQ to that<BR>
class, that radius bandwidth statement is like the equivalent of a<BR>
bandwidth statement under a normal physical interface, it now becomes<BR>
the global bandwidth for that virtual-interface that the router<BR>
recognises, you just have to make sure you do shape it properly though<BR>
otherwise your QoS won't be very affective.<BR>
<BR>
Cheers<BR>
<BR>
David Freedman wrote:<BR>
><BR>
> Thats great, the percentage I was referring to though was in the LLQ<BR>
> (as in priority percent and bandwidth percent) are these possible?<BR>
><BR>
> Dave.<BR>
><BR>
> ------------------------------------------------<BR>
> David Freedman<BR>
> Group Network Engineering<BR>
> Claranet Limited<BR>
> <A HREF="http://www.clara.net">http://www.clara.net</A><BR>
><BR>
><BR>
><BR>
> -----Original Message-----<BR>
> From: Ben Steele [<A HREF="mailto:ben@internode.com.au">mailto:ben@internode.com.au</A>]<BR>
> Sent: Wed 1/9/2008 23:53<BR>
> To: David Freedman<BR>
> Cc: cisco-nas@puck.nether.net; cisco-bba@puck.nether.net<BR>
> Subject: Re: [cisco-nas] Support for "percent" directive in<BR>
> per-session QoS<BR>
><BR>
> You specify your bandwidth via a radius av-pair, say for example you<BR>
> have a 512kb downstream ADSL service landing on your LNS you would have<BR>
> a line like:<BR>
><BR>
> cisco-avpair = "lcp:interface-config#1=bandwidth 435" (435kb<BR>
> allowing for a 15% dsl overhead)<BR>
> cisco-avpair = "lcp:interface-config#2=service-policy output SHAPED-QOS"<BR>
><BR>
> Then using the shape average percent command the LNS would shape the<BR>
> user to that speed, you can then apply your actual QoS policy under<BR>
> that, like so:<BR>
><BR>
> policy-map SHAPED-QOS<BR>
> class class-default<BR>
> shape average percent 100 50 ms<BR>
> service-policy QOS<BR>
><BR>
> This would apply the policy "QOS" to the shaped interface, I hope that<BR>
> is what you were after, and yes it does work quite well with<BR>
> priority(LLQ), the most important thing is to shape it to the correct<BR>
> speed, when the link is being saturated and you do a "sh policy-map int<BR>
> xx" you want to see it reporting that it is being shaped.<BR>
><BR>
> Cheers<BR>
><BR>
> Ben<BR>
><BR>
><BR>
> David Freedman wrote:<BR>
> > -----BEGIN PGP SIGNED MESSAGE-----<BR>
> > Hash: SHA1<BR>
> ><BR>
> > I believe this is possible reading the documentation at<BR>
> ><BR>
> <A HREF="http://www.cisco.com/en/US/products/ps6566/products_feature_guide09186a0080610dad.html">http://www.cisco.com/en/US/products/ps6566/products_feature_guide09186a0080610dad.html</A><BR>
> ><BR>
> > there is mention of it, but the question I want to ask is :<BR>
> ><BR>
> > how does it know what the available bandwidth is such to be able to<BR>
> > calculate the percentage?<BR>
> ><BR>
> > I mean, with a standard CBWFQ attached to an interface, it derrives this<BR>
> > from the interface bandwidth, but for vaccess QoS we are required to<BR>
> > nest the CBWFQ under a shaper, so does the percentage directive<BR>
> > use the shape average rate of this shaper?<BR>
> ><BR>
> > Does anybody use this directive (interested for those whom use it for a<BR>
> > priority queue) and does it work?<BR>
> ><BR>
> > Many thanks,<BR>
> ><BR>
> ><BR>
> > - --<BR>
> > David Freedman<BR>
> > Group Network Engineering<BR>
> > Claranet Limited<BR>
> > <A HREF="http://www.clara.net">http://www.clara.net</A><BR>
> > -----BEGIN PGP SIGNATURE-----<BR>
> > Version: GnuPG v1.4.2.2 (GNU/Linux)<BR>
> > Comment: Using GnuPG with Mozilla - <A HREF="http://enigmail.mozdev.org">http://enigmail.mozdev.org</A><BR>
> ><BR>
> > iD8DBQFHhRGktFWeqpgEZrIRAh4TAKDM9Kl+ikOSt8j19/9GfCDaTG8EegCgpYBW<BR>
> > AwNWZohx+5dbeFZ3pQkDr7A=<BR>
> > =Rn2w<BR>
> > -----END PGP SIGNATURE-----<BR>
> > _______________________________________________<BR>
> > cisco-nas mailing list<BR>
> > cisco-nas@puck.nether.net<BR>
> > <A HREF="https://puck.nether.net/mailman/listinfo/cisco-nas">https://puck.nether.net/mailman/listinfo/cisco-nas</A><BR>
> > <BR>
><BR>
><BR>
<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>