[c-nsp] Limiting BW in an HSRP set-up

[Oliver Boehmer (oboehmer)]

> please consider the following scenario :
> 
> *	typical datacenter with two routers running HSRP
> *	datacenter customer with 2 servers, each being connected on one
> dedicated (managed) ethernet port
> 
> How do I limit bandwidth for this customer?
> 
> I can easily limit upstream traffic, as the customer will probably
> use the HSRP address as a default GW; hence I can rate-limit on the
> .1q subinterface input.
> 
> But what about downstream traffic? I have the feeling that all I can
> do is to traffic-shape (or rate-limit) output on both .1q
> subinterfaces, but customer might then theoretically use twice the
> bandwidth he is paying for? 
> 

I am not sure if there is an easy technical solution for this problem,
you need a central point to enforce the policy. This can be a router
(i.e. a single link with a policer) or the backend/accounting system
where you add the used bandwidth on both links and bill them more if
they use more than they bought..

	oli