From: Vincent Mashburn (vmashburn@fedex.com)
Date: Mon Dec 12 2005 - 11:37:03 GMT-3
I am not sure that I agree with the statement from the book. With
CBWFQ, you can configure a class to match on the precedence or DSCP. To
my knowledge, there is no queuing that is "required' for a configured
TOS to work. However, during times of congestion, the higher priority
packets may me dropped due to the fact that the hardware queue (tx-ring)
is fifo. So, if you implement the queuing mechanisms stated in your
e-mail, your high priority packets will have a higher probability if
getting transmitted during times of congestion. Hope this helps.
Vince Mashburn
Engineer
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
san
Sent: Monday, December 12, 2005 1:04 AM
To: lab
Subject: TOS bits versus Queueing question
Quote from book: CCIE practical studies, vol 2, Page 189
Cisco IOS software considers the precedence bits of the TOS field if
there is traffic that is queued in WFQ, WRED, or WRR (weighted round
robbin). The precedence bits are not considered when policy routing,
PQ, CQ, CBWFQ are configured.
How about DTRC bits (bits 3 - 6) of TOS field,. Does this have any
limitation like above ?
After few pages, Page 192, example 2-31.
Example sets "set ip tos min-delay", & says. In this model, because
you are setting TOS values, you need to configure WRED or WFQ on the
outbound interface. WFQ is not the default queuing method on Ethernet
interfaces. It is the default queuing method on serial interfaces
with 2.048 Mbps or less of bandwidth. This portion of the
configuration is not present in this example.
Example 2-31.. shows configuring PQ though (&says PQ for TOS
enforcement on sidenote). ....Can anyone check, if the example is
correct. If example is correct then DTRC bits can have different
queue requirements than Precedence bits ? Is this correct ?
Summary: (To act on the bits)
Precedence bits => needs WFQ, WRR, WRED on outbound interface + queue
congestion
DTRC bits => any queue could work !?!
Thanks
SAN
This archive was generated by hypermail 2.1.4 : Mon Jan 09 2006 - 07:07:51 GMT-3