Showing results for 
Search instead for 
Did you mean: 
Walkthrough Wednesdays

QoS Opinions for Frame Network with Voice & Video

Hello, I have a config I am putting together for a customer who is running both voice and video over their frame network. The frame network consists of three sites; MI, DC, TX, with MI being the hub. The frame network is a full mesh with every site having a PVC to each other. Each site has a T1 connection rate with a full line rate CIR.

The challenges I'm having are how to tweak the QoS policies for both voice and video conferencing and also how to allow for the fact that the hub site (MI) has the same line rate (T1) as the other spoke sites. Obviously their is the potential for the spokes (DC & TX) to overrun the hub site, especially when we're talking a video conference between all three sites.

I was planning on implementing LLQ with FR Traffic Shaping but I'm indecisive about the bandwidth numbers I'm defining in the policy map classes.

The video conf. is polycom and I can control what DSCP/COS values the units tag at and also how much bandwidth they configure a session to, which should help. I planned on riding the voice in the priority queue and letting the video ride in a high-level queue with a COS of 4.

I'm interested to hear how others have configured their policies with regards to bandwidth when you have the same line rate everywhere.

Also anyone who has some suggestions on approaching this differently are more then welcome to comment. Sample configs would be great!

Thanks for the help!

Rising star

Highly recommend you review the QoS SRND. It will cover video, voice and data. Based off this should be running video in your LLQ. You can specifiy a 2nd LLQ statement (not a 2nd priority queue). This way you can police the Video traffic down if need be. Of course Cisco recommends that you never exceed 33% for your LLQ but I've been in shops doing upwards to 40% and still with little impact. If you have access to Netflow stats I've found this is the best way to size your queues and also restrict other unwanted traffic.

As far as shaping you aren't going to want to do that with Voice. Allocate enough voice bandwidth for your expected sessions. You are probably gonna want to look at IOS Gatekeepers for CAC as this will handle overall bandwidth to the site. Match this to your LLQ. CallManager locations CAC work more in a hub-and-spoke model. Your sites are meshed so this could be an issue for CM out of the box. Of course look at CM features like AAR, Regions with Low Bandwidth codecs, etc. in order to minimize voice bandwidth.

Here is the SRND link.

Please rate the posts



Content for Community-Ad

Spotlight Awards 2021