Options
QOS Scenario: Multisite QOS Design and Implementation
Hey TE,
Recently I switched a client to new MPLS circuits (from a consultant viewpoint, not ISP), and they've been having issues with the pipe bouncing all over the place
This is especially painful because they have VOIP going across these links and some mission critical items that need to be accounted for. Also, all internet-bound traffic goes over these links. A proxy server is being used to help with the web bandwidth, but there's still a lot of things going over a single T1. The deal is that ALL ROUTERS in this scenario are managed by the client's carrier. We take care of all their Layer 2/3 switching and VOIP. Now, we have arranged with the carrier that they are to trust our QOS marking across the MPLS domain, but my question is how will I configure the head-end site to not overflow the branch sites?
Here's the topology:
From my standpoint, it would be one thing if this was frame relay with a point to point configuration, but would I need to define classes for each site and police them down? The branches I could just do a standard CBWFQ/LLQ, but for the HQ to Branch communication - if left to a standard configuration it would see that it has 10Mbps of bandwidth and never worry about throttling it before putting on the line.
Hopefully someone with a bit more real world experience can enlighten me.
Thanks!
Recently I switched a client to new MPLS circuits (from a consultant viewpoint, not ISP), and they've been having issues with the pipe bouncing all over the place
This is especially painful because they have VOIP going across these links and some mission critical items that need to be accounted for. Also, all internet-bound traffic goes over these links. A proxy server is being used to help with the web bandwidth, but there's still a lot of things going over a single T1. The deal is that ALL ROUTERS in this scenario are managed by the client's carrier. We take care of all their Layer 2/3 switching and VOIP. Now, we have arranged with the carrier that they are to trust our QOS marking across the MPLS domain, but my question is how will I configure the head-end site to not overflow the branch sites?
Here's the topology:
From my standpoint, it would be one thing if this was frame relay with a point to point configuration, but would I need to define classes for each site and police them down? The branches I could just do a standard CBWFQ/LLQ, but for the HQ to Branch communication - if left to a standard configuration it would see that it has 10Mbps of bandwidth and never worry about throttling it before putting on the line.
Hopefully someone with a bit more real world experience can enlighten me.
Thanks!