Cisco Support Community
Community Member

rate limiting on port basis ...


i have seperate 2 routers with me .

i want to rate limit with different bandwidth ports such that when one link goes down the bandwidth requirement of the same port can be taken care by the othre live link .


i want to configure link A to use 10 MB and link B to use 15MB .when link A goes down the link B should be able to cater to Link B traffic also .its alike ,in a normal scenario both will use their respective bandwiths assigned ,but their limit capacity will increase only when their is a outage .

Thanks in advance


Re: rate limiting on port basis ...


From your post i feel you are asking about reduandcy in the first place and dynamic rate-limiting based on the traffic requirements when one of the links is down.

Do revert whether you want to deploy this using 2 different hardware platforms/routers or in a single router ?

AFAIK if you are having specific subnets using these links then you can make use of source based routing for deciding the primary and secondary path to be used.

Also you can match up the traffic source from different subnets using ACLs and have them binded on the respective rate-limit command to limit the bandwidth usage based on the subnet criteria..

if this doesnt solve ur purpose do throw some lite on ur query and the exact requirement u hve in out there..


Community Member

Re: rate limiting on port basis ...

thanku premkumar.

yes im looking for a dynamic rate-limiting kinda scenario.

i am planning to deply it on 2 seperate routers .i have tried it using ACL but it doesnt ensure the movement of traffic in normal basis.

like :at any point of time , traffic from SiteA should use 10 MB from certain link only but when i rate limit the same on other router ...this way i am not ensuring that same traffic traversing through A wont come here.same goes for Site B also.

kindly suggest..

Re: rate limiting on port basis ...


you should understand some limitations of the current QoS approach - DiffServ. In DiffServ what we do is to give a rule set to a router on how traffic across an interface in a specific direction should be treated.

The interesting traffic, which should be treated in a special manner has to be described to the router. This is called classification.

The question in your case is how to describe the traffic normally sent across router A and the traffic normally sent over router B.

What you can do (?) is to mark traffic normally sent over router A with precedence 1 and traffic normally sent over router B with IP precedence 2. And then on the output interface on BOTH routers you specify, that prec 1 is limited to 10 MB and prec 2 is limited to 15 MB. The point here is, in case there is no traffic of prec 1 across the interface it will not block any resources.

Example config:

class-map match-all RouterA

match ip precedence 1

class-map match-all RouterB

match ip precedence 2

policy-map OUT

class RouterA

shape 10000000

class RouterB

shape 15000000

class class-default



interface Ethernet0


service-policy output OUT

This will rate-limit your traffic to 10 MB or 15 MB respectively. The assumption for this example is, that the IP packets are properly marked in another place.

Hope this helps


Community Member

Re: rate limiting on port basis ...

thanks martin.

actually my setup is :

i have got 2 routers RA and RB connected to 2 sepearte locations L1 and L2.RA and RB are connected to each other also.

i want to rate limit my traffic from RA to L1 for 10MB .

i want to rate limit traffic from RB to L2 for 15MB.

INCASE ,of failure of link between RA and L1 i can route traffic from RB (as they are internally connected )i need to have rate limiting for this traffic from RA also to be set on RB to L2 .only main criteria is that this should only be applicable when the link between RA and L1 goes down else not .(rate limit to be set to 15MB only)

i hope im able to explain my need .

thanks in advance

Re: rate limiting on port basis ...


this is what the above configuration will do (assuming proper marking of IP precedence). It should be applied in RA and in RB!

The trick is f.e., that class routerA in the service-policy is only involved when there is matching traffic. This will only happen in RB INCASE there is a failure in the link between RA and L1.

Through normal IP routing you should ensure, that traffic to L1 only travels through RA and to L" only through Rb. Am I right that this is the case already?

Then the above does what you would like, as far as I understand your requirements now.

Hope this helps! Please rate all posts


Community Member

Re: rate limiting on port basis ...

hi martin,

ya thats correct.

actually i need to provide this for a customer and we have to ensure that by no chance the port gets overused at any point of time .


he doesnt over uses both the links at the same time without any outage/failure.

thanks again

Re: rate limiting on port basis ...


with the configuration above the maximum amount of traffic allowed will be 25 MB in any case, for no WAN failure, link to L1 down or link to L2 down.

So I still think the above configuration concept will fulfil your requirements keeping in mind the prerequisits stated.

Hope this helps!


Community Member

Re: rate limiting on port basis ...

thanks again.

need to know is it possible that same can be configured on dynamic basis .15MB only from RB when RA is Ok .but gets "25MB only when RA goes down" .

its redundancy sort of scenario .....


Re: rate limiting on port basis ...


this is not possible with routers allone. What you want is a configuration change based on an event happening in another router.

This is the reason, why I suggested a configuration supporting both cases.

What you could do is setting up a management station triggering a reconfiguration of RB, when a link failure in RA is detected (f.e. through a SNMP trap).

Hope this helps! Please rate all posts.


Community Member

Re: rate limiting on port basis ...


do you have the configuration for it .or could you propose something on it how to do it ???how do i configure that it get triggers...???

Thanks in advance

Re: rate limiting on port basis ...


this is a Unix scripting/programming task.

The trigger can be an SNMP trap or a syslog message sent by the router RB. In the end a rather complicated task, if you ask me. That is why I suggested the configs given.

Hope this helps! PLease rate all posts!

Regards, Martin

CreatePlease to create content