RE: RSVP and LLQ

From: Chris Lewis \(chrlewis\) (chrlewis@cisco.com)
Date: Fri Aug 19 2005 - 18:24:25 GMT-3


Gladston,

I find use of the term LLQ difficult to quantify exactly. LLQ came in to
being as a marketing term to describe the use of a priority queue
combined with policiing to support latency guarantees (a priority queue
by itself cannot do this as if an interface is congested with just a
priority queue, that is the same as having no queueing at all). From
there, LLQ got associated in most people's mind with a priority queue.

However, if you look at the essence of the two quotes, they are saying
the same thing. Assuming the interface supports it, there will typically
be one priority queue on an interface (you can configure more but that
is fairly meaningless from a practical perspective as you can only
transmit one packet on a wire at a time), and you have to configure a
way for a packet to gain admission to that priority queue. This can
either be done by configuring diff serv admission criteria (matching in
a class map and configuring class actions under a policy map), or you
can have RSVP admit traffic in to that queue.

This is true whether the interface is frame relay encapsulation at layer
2 or not.

Chris

________________________________

From: gladston@br.ibm.com [mailto:gladston@br.ibm.com]
Sent: Friday, August 19, 2005 1:17 PM
To: Chris Lewis (chrlewis)
Cc: ccielab@groupstudy.com
Subject: RE: RSVP and LLQ

Thanks a lot Chris,

Would you say that what you said...

==================
quoted
Flows within the RSVP traffic specification (TSpec) you setup via the
global pq-profile command use the interface priority
=================

...is different from the following Wendell's phrase?

====================
RSVP put voice-like traffic into the existing LLQ priority queue on the
interface
====================

When I read Wendell first time I thought it was refering to the queue
created by priority command under CBWFQ. (as soon as I read LLQ I
related it with MQC LLQ)

Now I am considering that it means the priority queue under the
interface, created by 'frame-relay fragment...'

------------------------------------------------------------------
Gladston

"Chris Lewis \(chrlewis\)" <chrlewis@cisco.com>

19/08/2005 12:07

To
Alaerte Gladston Vidali/Brazil/IBM@IBMBR, <ccielab@groupstudy.com>
cc
Subject
RE: RSVP and LLQ

Hi,

It looks like you are quoting
http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122cgcr/
fqos_c/fqcprt5/qcfrsllq.htm#1002796 it is correct I believe in terms of
the basic global and interface level configurations required.

Flows within the RSVP traffic specification (TSpec) you setup via the
global pq-profile command use the interface priority queue with no
further MQC configuration required. Flows with a TSpec above the PQ
profile get a reserved queue within WFQ.

With the basic config you show from the DoC CD for global and interface
configurations a voice-like PQ profile is enabled by default. If this is
applied for example to int s3/0, you get 75% of the interface 1544 Kb
for this traffic. In practice you would probably modify the pq-profile
from the default figures

Router1#sho ip rsvp int s3/0
interface allocated i/f max flow max sub max
Se3/0 0 1158K 1158K 0

Some general comments on this that may help.

This is a different concept to the more usual diff serv mechanism of
classifying traffic based off DSCP or ports and placing them in a queue.
Here RSVP is handling admission control, or rather admission to the
priority queue. This is more of an edge rather than core mechanism. It
is useful for limiting the amount of PQ traffic allowed on a constrained
bandwidth access link. In the core of a network where thousands of flows
exist, per flow mechanisms break down and diff serv is more commonly
deployed.

Chris

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
gladston@br.ibm.com
Sent: Friday, August 19, 2005 7:42 AM
To: ccielab@groupstudy.com
Subject: RSVP and LLQ

Do you have a workable example of configuring RSVP support for LLQ?

Reading Wendell and Doc CD left some doubts.

Doc CD example:
global:
ip rsvp pq-profile

interface:
ip rsvp bandwidth
fair-queue

Wendell shows these commands on the example:
global:
ip rsvp pq-profile
ip rsvp bandwidth

interface:
service-policy...

I understood the concept explained under Wendell, but it is not clear
how to configure it.

Any feedback appreciated.



This archive was generated by hypermail 2.1.4 : Sun Sep 04 2005 - 17:01:19 GMT-3