Real CBWFQ problem

From: Serge (serge.ciscostudy@gmail.com)
Date: Mon Nov 26 2007 - 04:24:22 ART


Hi all!

I have the following situation:
policy-map SHAPE-OUT-DIVISION
  class WU-OUT
  bandwidth percent 5
  class MAIL-OUT
  bandwidth percent 15
  class PROXY-OUT
  bandwidth percent 5
  class HTTP-OUT
  bandwidth percent 2
  class SSH-OUT
  bandwidth percent 3
  class CYTRIX-OUT
  bandwidth percent 60
  class ALL_OTHER-OUT
  bandwidth percent 10
policy-map GRE128-DIVISION
  class ALL
  shape average 120000
  service-policy SHAPE-OUT-DIVISION
  class class-default

interface Tunnel5014
 description
 bandwidth 128
 service-policy output GRE128-DIVISION

and my client says that when somebody sends mail (traffic goes to the class
MAIL-OUT), CYTRIX users are suffering large delays. Probably, mail traffic
gets all the bandwidth and cytrix users don't get the reserved 60 kbit. It
seems like the priority 60 can help for the CYTRIX class, but then it will
be policed all the time (given the condition that the bandwidth is only 128
kbit, I assume that interface is congested all the time).

What can we do in such a situtation?



This archive was generated by hypermail 2.1.4 : Sat Dec 01 2007 - 06:37:31 ART