Hi,
I was hoping you could help with an issue one of our customers is having with Microsoft Teams. Since the lockdown they have been using teams more and more. The problem is when they have a video call with more than 5 participants the quality of the video is really bad and sometimes audio. I though it may be the people they are connecting to at home but it seems to be only the users at the customer premises that are affected. They have the call quality facility enabled and it shows a maximum of 90% packet loss. They have a 1gbps link which on which they only use around 100mbps. This goes across our 10 gbps mpls backbone and then to their firewall at the other side of our backbone which has a 1gbps connection to the internet. The firewall only gets to around 150mpbs at peak times
The customer originally had an srx 210 which has an mpls connection onto our backbone and they break out to the internet with a fortigate firewall. I have replaced this with an srx 340 cluster and done some work to enable cos on their router and our backbone which has made things better but its still not working great. Before I enabled cos i could see dropped packets on the interface going to their LAN which is made up of cisco switches. This seems to have stopped but the perfromace isn;t much better.
The customer has qos enabled on their cisco switches and if I do a packet capture I can see the right dscp code points are passed for the ports
I have attached the site cos and network (site router.txt). Its set to use the dscp values from the customer with the following values for teams
Audio 50,000–50,019 TCP/UDP 46 Expedited Forwarding (EF)
Video 50,020–50,039 TCP/UDP 34 Assured Forwarding (AF41)
Application/Screen Sharing 50,040–50,059 TCP/UDP 18 Assured Forwarding (AF21)
I have changed buffer size on the ques but doesn't seem to be making much difference. I have also replicated these settings across our backbone with the samer configuration. (durham exchange backbone.txt)
The call quality shows its outbound that seems to be the problem The customer said they think it may have been inbound before I made the changes but I cannot be certain on this (call quality.jpg)
I'm pretty sure its not the firewall I've done various tests over the last few weeks and if I connect straigh to the firewall I can make calls fine with any number of users connected. I don't think its the customers LAN setup either. Only the old router I gave them another port on their routing instance to plug a laptop directly in and it had the same issues when a number of people were on the call
I'm thinking its my cos setup as I've had very little experience of it before this issues. Does my setup look ok for use with Teams? I'm not sure if I possibly needed the classifiers and schedulers on both ingress an egress on their router and right accross the backbone so that cos would be applied in both directions
If the cos looks ok does anyone have any other ideas what it might be? I've been working on this for weeks now and the customer is running out of patience
I've also attached the que counters on their router (que counters lan.txt, que counters wan.txt)and finally the cos setup on their cisco switches (cisco.txt)
Any help would be greatly appreciated