Configuring QoS for priority queueing and bandwidth reservation
Hi folks, can someone explain how to set up QoS on a BIG-IP to get priority queueing and bandwidth reservation working for packets that are already QoS tagged (L2 CoS, L3 DSCP)? The scenario is the following: Network packets will be marked by the endpoints (or other devices) on Layer 2 (CoS) and Layer 3 (DSCP) according to the following scheme: That being said the packets that enter the BIG-IP already carry their QoS tags. The BIG-IP shall do two things with the packets: The QoS tags must not be changed and remain their original values For every application/traffic class the BIG-IP shall reserve a specific bandwith and assign a priority queue according to the QoS class On the switches and routers the following QoS polices have been implemented which ideally should be the same for the BIG-IP: For the first point (leaving the QoS tags unchanged) I think the solution is to configure the UDP/TCP profiles with IP ToS/Link QoS: Pass Through But how do I accomplish the second point (bandwidth reservation and priority queueing)? In the Network tab I found Class of Service and Rate Shaping, in the Acceleration tab there is Bandwidth Controllers. Which feature is the right one and how does it need to be configured for this scenario? Thx a lot!
Traffic Shaping outbound do the client from APM
So I know and have implemented in a lab traffic shaping from the VPN client to the APM server, which works quite well and should shape large file-transfers inbound. Is there a way to do the opposite, i.e. traffic-shape to the client so they don't go pulling down large files from the corporate network and effectively consume the internet bandwidth? so I would like to say you user-group A can have 5Mbps, you user-group B can have 10Mbps etc.. cheers!ToS getting reset to 0 when egress from LTM
Scenario: Micrsoft Lync front-end servers. Their gateway is on the LTM. Client requests to the front-end are balanced by virtual servers on a variety of ports. The Lync servers are configured to mark their SSL-TLS SIP traffic (port 5061) as AF31. From packet captures, I have found that the markings are in tact while traversing the link to the LTM back-end network. However, when getting capture from the front-end interfaces, the DSCP has been reset to 0. I have a TCP profile set on the virtual server that is listening on 5061, which has the options enabled to pass-through ToS and QoS. What else am I missing here? I found a bug that looks somewhat related, but I am not running a SIP profile and this traffic is TCP. http://support.f5.com/kb/en-us/solutions/public/14000/000/sol14019.html
