RE: WRED in a Subinterface

From: Scott Morris (swm@emanon.com)
Date: Sun Jul 30 2006 - 16:48:37 ART


You are implying that logic follows all programming!

Ummmmm... Just because that's the way it is? I'm sure (I'll pretend
anyway) there's a good programming explanation for the way things change
based on those commands and in what order you put them in, but heck if I
know what they are!

Scott

-----Original Message-----
From: Victor Cappuccio [mailto:cvictor@protokolgroup.com]
Sent: Sunday, July 30, 2006 2:31 PM
To: swm@emanon.com; 'Schulz, Dave'
Cc: ccielab@groupstudy.com
Subject: RE: WRED in a Subinterface

Scott, another question for the record :)))

I must admit QOS Rocks not for cool, but for confusing.

R#show int s0/0 | in Queu
  Queueing strategy: weighted fair
WFQ is enabled.
now if

R6#conf te
Enter configuration commands, one per line. End with CNTL/Z.
R(config)#int s0/0
R(config-if)#random-detect
R(config-if)#do show int s0/0 | in Queu
  Queueing strategy: random early detection(RED)

It changes the queuing Name. This is nice, because it adds more confusion to
the whole thing.

Now, AFAIK, WRED is a Queue Cleaner discipline to say It in that way that
flushes depending on the average queue the TOS Value of the Packet based on
a minimum and maximum threshold, and is not a selective buffer for weighting
fair dynamic Queue for Flow.

But..

R(config)#int s0/0
R(config-if)#fair
Must remove RED configuration first.

So this is nice, Which Scheduling Strategy is using the interface now?
WRED??

And to add more confusion to the game

R#conf te
Enter configuration commands, one per line. End with CNTL/Z.
R(config)#policy-map TEST
R(config-pmap)#class class-default
R(config-pmap-c)#random
fair-queue or bandwidth on the class is required to issue this command

Could you please explain this like for a 10 year old kid?
Thanks again.
Victor.-

-----Mensaje original-----
De: Scott Morris [mailto:swm@emanon.com] Enviado el: Domingo, 30 de Julio de
2006 10:12 a.m.
Para: 'Schulz, Dave'; 'Victor Cappuccio'; ccielab@groupstudy.com
Asunto: RE: WRED in a Subinterface

While that would make logical sense, it has been my experience that the
policy has to fully apply before the shaping takes place, and therefore the
answer would be no. You must to the child policy.

Unless, of course, they change the programming behavior. But if they're
going to do that, I would like to hope they would make a "let me make a
queue for you since one doesn't exist" feature. :)

 
Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, JNCIE
#153, CISSP, et al.
CCSI/JNCI-M/JNCI-J
IPExpert VP - Curriculum Development
IPExpert Sr. Technical Instructor
smorris@ipexpert.com
http://www.ipexpert.com
 
 

-----Original Message-----
From: Schulz, Dave [mailto:DSchulz@dpsciences.com]
Sent: Sunday, July 30, 2006 1:13 AM
To: Scott Morris; Victor Cappuccio; ccielab@groupstudy.com
Subject: RE: WRED in a Subinterface

Scott -

Can I assume that we could use the shape and random-detect in the same
policy-map, and thereby eliminating the need for a child policy? Like
this....

policy-map TOP-LEVEL
 class class-default
  shape average 64000
  fair-queue
  random-detect

Dave Schulz,
Email: dschulz@dpsciences.com

-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Scott Morris
Sent: Saturday, July 29, 2006 11:33 PM
To: 'Victor Cappuccio'; ccielab@groupstudy.com
Subject: RE: WRED in a Subinterface

Queues don't exist on subinterfaces, therefore any queuing functions can't
be done. This is true of any sub-interface (serial, ethernet, etc.)

When you employ shaping, however, you create a queue. Once you have a
queue, you can using it for queuing.

Policy-map WRED
 class class-default
  random-detect

Policy-map TOP-LEVEL
 class class-default
  shape average 64000
  service-policy WRED

Int s0/0.15
 service-policy output TOP-LEVEL

HTH,

 
Scott Morris, CCIE4 (R&S/ISP-Dial/Security/Service Provider) #4713, JNCIE
#153, CISSP, et al.
CCSI/JNCI-M/JNCI-J
IPExpert VP - Curriculum Development
IPExpert Sr. Technical Instructor
smorris@ipexpert.com
http://www.ipexpert.com



This archive was generated by hypermail 2.1.4 : Tue Aug 01 2006 - 07:13:48 ART