I have one primary and one secondary OC3 at my hub along with full T1's and DS3 connections at my remote sites. At my hub site I currently have two 6509’s with pos cards in them running a sup-720. I currently have 83 PVC's hanging off my one OC3 interface connection (running FIFO). Along with that all the data is being encrypted via ipsec due to its classified information. I also have GRE tunnels running but that is just for my IPX traffic so we don’t have to worry about those tunnels. All the agency routers have vpn accelerator cards in them. Now for the fun part I need to run video across to 11 sub-interfaces to my remotes sites and guarantee them 340K bandwidth and the rest can be fair-que. On my DS3 sub-interfaces I rate-limit them to 9 Megs but nothing on my T1's. Quick note also we have no cir value set from the provider, they give us a full 1.5 Meg to each site at layer 2. They are running a MPLS network with a martini draft vpn between us. When video traffic is not being used Data needs full access to the link since some of the T1's are running at 75% right now with no video. We need to be able to do this on a sub-interface level and restrict the Hub from send more the 1.5 Megs of data out the sub-interface while trying to send Video. I looked into LLQ for frame-relay but it looks like that you have to turn on frame relay traffic shaping on for the main interface and then create map classes along with your class maps and policy maps. The problem with that is it affects my complete OC3 with traffic shaping since the documents says that it will use default values for traffic shaping if you don't classify them per sub-interface. I need a solution that I can just apply to 11 sub-interfaces and then on the remote side I can apply the same thing. Any help on this solution would be greatly appreciated. I looked at several documents but can't seem to find one that will work in my environment. My congestion is inbound to my remote sites coming from the hub.
Thanks,
John
