01-09-2009 07:49 AM - edited 03-04-2019 03:24 AM
I have an interesting scenario that I am working on putting together a solution for.
I have a metro Ethernet network with only a Gigabit backbone connection between 2 locations. I need to give a customer a 145Mbps circuit using a gige handoff on both ends, but do not want to give them the ability to use the full gig of course.
Now I know this is easily done with things like port channels and load balanced links, but the major factor in this situation is that there is a large amount of VoIP traffic being sent over the link(s).
I thought about doing some policy maps and traffic shaping down to 145mbps, but then I figured that the buffering of traffic when the customer is trying to exceed that 145mbps may cause issues with voip packets and latency.
Again, the issue with traditional load balancing with routing protocols is the possibility of packets arriving out of order etc.
I was looking toward a way to shape the traffic, as well as have all Voip traffic have priority and never be buffered, but if all traffic was voip, and the customer was trying to use say 150mpbs, then some would always be queue and then their would be latency?
Thanks for any response.
01-09-2009 12:10 PM
Hello Jason,
your scenario could receive a good answer from hierarchical Modular QoS where two policy maps are used:
one to shape all traffic to 145 Mbps
and one to provide to VOIP traffic a LLQ treatment within it.
see for example
the config should be something like:
Interface gi0/0
Service-policy output parent
policy-map parent
class class-default
shape average 145000000
service-policy child
policy-map child
class voice
priority 50000000
class video
bandwidth 6000
class data
bandwidth 3000
Hope to help
Giuseppe
01-09-2009 12:50 PM
This does help, I was thinking it would be something with policy and QoS, but I was just not sure on the exacts.
Thank you for the response. I will try this out.
01-10-2009 05:32 AM
Just wanted to further comment on the difference between what Giuseppe is suggesting and what I'm suggesting. Giuseppe's config would be the method of choice if the WAN was limited to 145 Mbps but had higher bandwidth edge handoffs. (e.g. Ethernet gig handoffs but OC-3 in between.) My config assumes the WAN is the gig you noted but you want to restrict a customer's "slice" of the bandwidth. If the latter, Giuseppe's config, although it places customer VoIP first within the shaper's control doesn't guarantee VoIP traffic not under control of the shaper, and if there is traffic outside of the shaper, would not really guarantee customer's VoIP perfomance. It very important to understand whether you desire to control all traffic limited by a downstream bottleneck or just the customer's traffic.
01-09-2009 12:28 PM
From what I believe you're describing, you have a gig metro link that has VoIP and you want to allow a customer 145 Mbps on it, something like this should work:
policy-map gigE
class yourVoIP
priority xxx
class yourTraffic
bandwidth xxxx
class yourCustomer
bandwidth 145000
shape average 145000000
If your customer also has VoIP, then you should define a LLQ class for it (you might deduct the bandwidth for it from their overall allocation).
e.g.
policy-map gigE
class yourVoIP
priority xxx
class customerVoIP
priority 5000
class yourTraffic
bandwidth xxxx
class yourCustomer
bandwidth 140000
shape average 140000000
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide