QoS Question

Unanswered Question
Jul 30th, 2009
User Badges:
  • Bronze, 100 points or more

Hello,

I have what is hopefully an easy question for someone to answer regarding QoS.


Firstly, If I understand it correctly, QoS only kicks in when congestion is experienced on a link. How does the switch / router determine that a link is congested? Do you have any good links that could explain this better for me?


Secondly, What I'm trying to accomplish is to have some replication traffic across the WAN be allowed to use as much bandwidth as it wants up until there's congestion, then i want that traffic to be the first traffic to be dropped to allow other business traffic / VoIP to pass. How would I go about this?


Platform is 4507R with SupIV


QoS is honestly a very weak point with me. Thanks in advance for any help.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Joseph W. Doherty Thu, 07/30/2009 - 11:47
User Badges:
  • Super Bronze, 10000 points or more

Much QoS does "kick in" when there's congestion, but not all. The QoS that treats congestion, generally, is queue management. (I.e. congestion is indicated by packets/frames queued.)


How you accomplish your policy of managing VoIP, ordinary business traffic and background depends on both the nature of the circuit and the QoS features of the platform.


Many LAN oriented switches tend to be feature poor for QoS. For the 4500 SupIV, there are 4 transmit queues per port, but feature restricted to some ports.


I.e.


"For systems using Supervisor Engine II-Plus, Supervisor Engine II-Plus TS, Supervisor Engine III, and Supervisor Engine IV, bandwidth can be configured on these ports only:


•Uplink ports on supervisor engines


•Ports on the WS-X4306-GB GBIC module


•Ports on the WS-X4506-GB-T CSFP module


•The 2 1000BASE-X ports on the WS-X4232-GB-RJ module


•The first 2 ports on the WS-X4418-GB module


•The two 1000BASE-X ports on the WS-X4412-2GB-TX module "


Assuming you are using one of the supported port types, and assuming the interface port bandwidth is the WAN bandwidth (often not the case), you might place VoIP traffic into Q3 configured for strict priority, replication traffic in a different dedicated queue with minimum bandwidth specified (e.g. Q1) and normal business traffic (e.g. Q2) in another dedicated queue with maximum bandwidth specified. (You'll have a queue left over that you can use later for other critical traffic.)


If your WAN path has less bandwidth than your port interface, and since the SupIV doesn't, I beleive, support queue bandwidth allocations within a shaped rate, I would recommend you drop another device in-line (even a 8 port 2960 would help).


More information can be found staring in the configuration guide for your IOS in the chapter "Configuring Quality of Service".


PS:

More QoS recommendations for 4500s:

http://www.cisco.com/en/US/docs/solutions/Enterprise/Video/tpqoscampus.html#wp1062354

http://www.cisco.com/en/US/docs/solutions/Enterprise/WAN_and_MAN/QoS_SRND_40/QoSCampus_40.html#wp1099634

http://www.cisco.com/en/US/docs/solutions/Enterprise/WAN_and_MAN/QoS_SRND/QoSDesign.html#wp999958

Actions

This Discussion