From: Ramasubramanian Sethuraman (snrmanian@hotmail.com)
Date: Fri Feb 27 2004 - 12:28:52 GMT-3
Scott,
Thanks for the reply.
I also read dscp-to-cos and i was understanding like this
case a ) trust dscp (or) use dscp-dscp mutation at the i/p interface that
results in foll flow for a packet inside 3550
dscp --- dscp --- policed dscp --- cos --- o/p queue selected
case b ) trust precedence (or) use ip-precedence-dscp map at the i/p
interface
prec --- dscp --- policed dscp --- cos --- o/p queue selected
case c ) trust cos (or) use cos-dscp map at i/p interface
cos --- dscp -- policed dscp -- cos --- o/p queue selected
In all above cases, 64 possible dscp's at the end of policing (if its
configured)
will be mapped to one of 8 possible cos values. Even though multiple dscp's
can be
mapped to same COS, one dscp(for eg dscp 8) can be mapped to more than one
COS.
In such a case, each dscp will be present in only queue.
Pls clarify if my above understanding is right.
So i was confused when i saw the statment
>As a result of this configuration, when the queues 1 and 3 are filled above
>50 percent, packets with DSCPs 0, 8, 16, 24, 32, 40, 48, and 56 are
>randomly
>dropped. The same packets are randomly dropped when queues 2 and 4 are
>filled above 70 percent. When the second threshold (100 percent) is
>exceeded, all queues randomly drop packets with DSCPs 10, 20, 30, 40, 50,
>and 60.
The above implies that packets with same DSCP can be spread across the 4
queues
and depending upon the threshold for each queue, these packets will be
random dropped.
For eg) packets with dscp 8 will be randomly dropped from queue 1 and 3
if their queues are above 50 percent and similarly packets with dscp 8 wil
be randomly
dropped from queues 2 and 4 if the queue is above 70 percent.
Pls clarify.
thanks,
Subbu
>From: "Scott Morris" <swm@emanon.com>
>To: "'Ramasubramanian Sethuraman'"
><snrmanian@hotmail.com>,<ccielab@groupstudy.com>
>Subject: RE: cat35550 qos
>Date: Fri, 27 Feb 2004 09:32:11 -0500
>MIME-Version: 1.0
>Received: from schlep.emanon.com ([208.158.37.22]) by mc3-f41.hotmail.com
>with Microsoft SMTPSVC(5.0.2195.6824); Fri, 27 Feb 2004 06:32:15 -0800
>Received: from [12.37.38.254] by schlep.emanon.com (GMS
>9.02.3180/NT1887.00.097ced12) with ESMTP id eogtcaaa for
>snrmanian@hotmail.com; Fri, 27 Feb 2004 09:12:38 -0500
>X-Message-Info: JGTYoYF78jGF2bKabx8IKazsNSdiYtGZ
>X-Mailer: Microsoft Office Outlook, Build 11.0.5510
>Thread-Index: AcP869Bgh5cm7DojTwqoI2CVV1NU1AAUbb8g
>X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165
>In-Reply-To: <BAY2-F137dH995ZFM28000091dc@hotmail.com>
>Message-Id: <14123767124181@schlep.emanon.com>
>X-AntiSpam: Checked for restricted content by Gordano's AntiSpam Software
>Return-Path: swm@emanon.com
>X-OriginalArrivalTime: 27 Feb 2004 14:32:15.0683 (UTC)
>FILETIME=[7EFD3930:01C3FD3E]
>
>There are also dscp-cos maps that you can change along the way. Mutation
>maps are interesting creatures to play with.
>
>Default operation is to map IP Prec directly to COS (each have 8 possible
>values). DSCP has 64 possible values, so there's a little overlap there.
>
>64 --> 8 --> 4 queues There has to be SOME mapping going on!
>
>Any of these commands allow you to start tweaking the default maps as well
>as being able to implement a policy regarding their setup and handling.
>
>The wrr-queue dscp-map option allows you to map particular DSCP values to
>one of two discard thresholds. So it's not so much that a DSCP value can
>exist in more than one place, but that you have less WRED maps than you do
>queues. Think of it like applying one policy to multiple interfaces.
>You're setting some values that won't exist in certain places, but it makes
>the overall configuration smaller and more compact.
>
>Check out
>http://www.cisco.com/en/US/tech/tk389/tk813/technologies_tech_note09186a0080
>1558cb.shtml#topic4 it has some decent discussions about the mapping.
>
>HTH,
>
>
>Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, CISSP,
>JNCIS, et al.
>IPExpert CCIE Program Manager
>IPExpert Sr. Technical Instructor
>swm@emanon.com/smorris@ipexpert.net
>http://www.ipexpert.net
>
>
>-----Original Message-----
>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
>Ramasubramanian Sethuraman
>Sent: Thursday, February 26, 2004 11:53 PM
>To: ccielab@groupstudy.com
>Cc: snrmanian@hotmail.com
>Subject: cat35550 qos
>
>I have a question regarding 3550 Qos. I read that based on the internal COS
>value of the packet, one of the 4 o/p queues is selected. But the COS value
>itself is based on the DSCP value of the packet after policiing, which
>implies there is a one-to-one correspondence between DSCP and the selected
>queue (each queue has unique set of DSCP's)
>
>The point i don't understand is there is also a option(ingress interface)
>to
>map dscp to either of the 2 thresholds for each o/p queue using the command
>"wrr-queue dscp-map 1|2 <dscp set>". This tells that same DSCP can exist on
>multiple queues which contradicts my earlier understanding that each queue
>has a unique set of DSCP's.
>
>Pls clarify if i'm missing something.
>
>This is the sample config from CCO
>
>Switch(config)# interface gigabitethernet0/1 <-- egress
>Switch(config-if)# wrr-queue random-detect max-threshold 1 50 100
>Switch(config-if)# wrr-queue random-detect max-threshold 2 70 100
>Switch(config-if)# wrr-queue random-detect max-threshold 3 50 100
>Switch(config-if)# wrr-queue random-detect max-threshold 4 70 100
>Switch(config-if)# exit Switch(config)# interface gigabitethernet0/2 <--
>ingress Switch(config-if)# mls qos trust dscp Switch(config-if)# wrr-queue
>dscp-map 1 0 8 16 24 32 40 48 56 Switch(config-if)# wrr-queue dscp-map 2 10
>20 30 40 50 60
>
>explanation
>---------------
>As a result of this configuration, when the queues 1 and 3 are filled above
>50 percent, packets with DSCPs 0, 8, 16, 24, 32, 40, 48, and 56 are
>randomly
>dropped. The same packets are randomly dropped when queues 2 and 4 are
>filled above 70 percent. When the second threshold (100 percent) is
>exceeded, all queues randomly drop packets with DSCPs 10, 20, 30, 40, 50,
>and 60.
>
>
>
>
>thanks,
>subbu
>
>_________________________________________________________________
>Oscar fever is on! It's Hollywood's biggest party.
>http://server1.msn.co.in/sp04/oscars/index.asp Get all the dope here.
>
>_______________________________________________________________________
>Please help support GroupStudy by purchasing your study materials from:
>http://shop.groupstudy.com
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
>
This archive was generated by hypermail 2.1.4 : Fri Mar 05 2004 - 07:13:59 GMT-3