Hierarchical QOS over two logical connection

[KAROLY KOHEGYI]

Hi!

We have two logical connection which are connected via  1 physical Ethernet interface to 2 routers in central sites.

Both connections are 2 Mb/s. How can i classify the output traffic in order to shape both direction to 2-2 Mb/s.

There are not suitable "match" command!!! ??

The branch routers are 2650xm and 2811.

Any help!

[KAROLY KOHEGYI]

Hi,

We must use service-policy on physcal interface because of VRF's.

So the service-policies on the subinterfaces are not useable. ( so its too easy )

[KAROLY KOHEGYI]

IT SEEMS TO BE INSOLVABLE PROBLEM WITH CISCO!

[Marwan ALshawi]

can you describe what is insolvable please ?

what is the device type that connect to the two routers ?

have you tried to use ACL with policing by matching any traffic going to anything behind router 1 to be policed to 2 M

and another class for any traffic going to networks behind router2 to be policed to 2 M

then put those classes in policy map and apply the service policy to the main interface

also you might try to match the next hope router MAC address in a class map and police the traffic to 2 M using the below match under the class map if your IOS support it

class-map xxx

match destination-address mac address

hope this help

if helpful rate

[KAROLY KOHEGYI]

Thanks your answer!

Thanks the destination-address MAC address match is good but not "elegant" in this case.

The real situation is the following.

Every branch has two routers. One of routers is the HSRP active and we use load-balancing.

Some places the HSRP active router is the 2650XM which has one Ethernet interface.

The output packets might have three direction. 1. Local LAN 2. R2 router 3. Central routers

The destination mac-address match is also good but I would like to use service-policy  which is SAME in every routers.

The router cannot distinguish the packets going out on the parameter basis of the packets.

The mac-address is unique so in this case we need to configure all routers one by one!

Very thanks!