Just a quick question. I have been getting more and more into Ip phone installation with IP office and different routers/switches
Do you make it a pratice of changing the diffServe settings depending on the switch your connecting with or leave them default?
Yes but QOS would be implemented at the router stage for ip trunk and at the switch port stage to prioritze queued traffic. The point is that you can change either the router and switch diffserve settings to match ipo or vice versa.
If qos for voip has not been implemented at the site then it is better to match the existing ipo settings.
Don't forget the default DSCP value is 46 which means EF (expedited forwarding) and is the most widepread setting on IP telephony for this (not just by Avaya) and will be honoured by most equipment by default
Thanks for the info and education.. Some of these settings I never touch until someone asks about them... The DSCP value for the Cisco that is connected is 26 and I have change the IP office to match that.. seems to be working fine..
That is just plain wrong. DSCP 26 is AF (Assured Forwarding) with low drop probability but still qualifies for drop. Voice and video traffic should by definition always be marked as EF (Expedited Forwarding) aka DSCP 46.
I would as well if I had both on the same network as voice is more important than video however when we do only video and no voice I mark it as EF to avoid the quality complaints.
I like a good discussion/education. So correct me if I am not right, But, basically if QOS is set on network I should just leave the IPO difserv setting at default..
There are standards though. Voice wants low delay, low jitter and no drop rate. This is the definition of Expedited Forwarding (EF / DSCP 46). Sure you can use other things. Hell for that matter you can map Best Effort (BE) traffic to your high priority COS queue and really do something funky.
QoS on switches are usually handled in one of two ways. Either they recognize and act on the raw DSCP values or they map DiffServ to Class of Service (COS) queues and each vendor had different numbers of COS queues in their various switches for more granularity. It is usually the mapping of DiffServ to COS where things get screwed up.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.