Why you should never route internal Skype4B media traffic externally.

This puzzles me.
I have on several occasions been called in to solve issues with varying audio- and video quality at customers with Lync 2013 and Skype for Business deployed on premise.
After som tracing of media flow I have found that some of these for some reason has chosen to route all internal sip traffic towards the external interface of the edge server.
This all would still work better if media was able to route peer to peer between the clients.
In many cases this is not so.
The internal networks does not allow this so media is routed through edge.
This causes severe jitter and increases round trip significantly.
Video will appear as lagging and audio will be distorted or clipping.
The Skype4B quality indicator in the client will turn orange or red indicating that something is wrong.
Furthermore some equipment like Polycom phones and similar will not enjoy getting a certificate form the external interface of the edge server.

So what I do is usually to change/add the internal DNS records so that the clients signs inn directly to the front end servers and not via edge.
Then there is the other part to ensure that media can be routed internally.
When this is changed your media quality will improve significantly.
If anyone experience a similar problem, check if this is the case.
I've found no good reason to why some customers choose to do so.
Maybe because they have multiple sip domains and wish to keep the DNS records to a minimum internally.
Any ideas to why this is done is greatly appreciated.

Comments

  1. I have seen that some customers don't accept P2P between clients and not traffic between clients and frontend server. Then the traffic has to go out on the internet, then back to Edge and to Frontend. Not great, but it works, but you can get trouble. :-)

    ReplyDelete

Post a Comment

Popular Posts