From: Dennis J. Hartmann (dennisjhartmann@hotmail.com)
Date: Mon May 09 2005 - 21:17:53 GMT-3
ONLY policing and shaping set ceilings (limits). The bandwidth and
priority commands create guaranteed mimimuns during congestion. If there's
no congestion, the traffic can spike up to interface bandwidth (unless
policing or shaping is configured). The priority queue guarantees delay,
packet loss, and delay variation (jitter), while the bandwidth statement
only guarantees packet loss. The priority queue is FIFO and is strictly
scheduled (up to the priority bandwidth configuration). The PQ can consume
the entire pipe size unless there's a policer or shaper.
-Dennis Hartmann
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
ccie2be
Sent: Monday, May 09, 2005 6:37 PM
To: gladston@br.ibm.com; ccielab@groupstudy.com
Subject: RE: Priority Queue on CBWFQ
I think Chin is correct.
My understanding is that when shaping is configured, the parameters are
floors or ceilings that apply ONLY DURING CONGESTION.
So, the priority queue will set a maximum DURING CONGESTION and the other
queues, configured with bandwidth, set a minimum DURING CONGESTION.
In addition, priority causes packets in that queue to go first DURING
CONGESTION.
If anyone thinks differently, please let me know.
HTH, Tim
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
gladston@br.ibm.com
Sent: Monday, May 09, 2005 12:58 PM
To: ccielab@groupstudy.com
Subject: Priority Queue on CBWFQ
Hi,
Chin's book, page 544 says:
"When the CBWFQ queues are not using up their allocated bandwidth, traffic
from a congested priority queue can use up the free bandwidth."
He follows with and example where a VIDEO class is configured for priority
64kbps and show frame pvc xxx shows the class using 86kbps.
I always considered priority police a traffic, so priority 64 would give it
stricty priority but also would police, so it can not go further than 64kbps
(no matter if there is congestion or not).
What do you think?
I have the impression that Chin's error is to consider that "offered rate
86000" means it is using the bandwidth, when in fact we can see drops packet
increasing and show int ser x/y shows 64kbps. At least on the practice lab I
did that is what occurs.
This archive was generated by hypermail 2.1.4 : Fri Jun 03 2005 - 10:11:57 GMT-3