QoS Control by Means of COPS to Support SIP - Based Applications

The COPS protocol has been designed to enable communication on the interface between the policy decision administrator and the policy enforcement devices in a policy-based networking environment. It can be recognized that on the same interface there is the need to transfer information related to the request of resource by QoS clients and for the allocation of resources by resource allocation servers (e.g., band-width broker) in a DiffServ network. Hence, it is sensible to add this resource allocation functionality in the COPS framework. In particular, there are at least two cases where it is sensible to use COPS. The first case is on the interface between an edge node and a resource control node for handling resource allocation in a network provider domain. The second case is on the interface between a customer (client of a QoS enabled network) and the network provider: here COPS can be used as a protocol to signal dynamic admission control requests. In this article we present the definition of a new COPS client type to support the above-mentioned functionality, then describe an application scenario where SIP-based IP telephony applications can use Diffserv-based QoS networks. Simple backward-compatible enhancements to SIP are needed to interact with COPS/Diffserv QoS. A testbed implementation of the proposed solutions is finally described. he Common Open Policy Service (COPS) protocol has been defined in the context of the IETF RAP working group as mean to support policy control in an IP quality of service (QoS) environment. The underlying architectural model foresees that policy servers administrate the network communicating decision to policy clients (e.g., network elements) where the policy decisions are enforced. Basically the decisions concern who is authorized to access what resource in the network. In particular, if IP QoS is deployed, the users can access different transport services, and this access must be administratively regulated. Two architectural models for IP QoS have been proposed in the Internet Engineering Task Force (IETF): the integrated services (IntServ) and differentiated service (DiffServ) archi-tectures. A very good introduction to the topic of IP QoS can be found in [1], where the two approaches are described and compared. Due to scalability issues of the IntServ model, the DiffServ model seems preferable for the development of IP QoS in a real-life network. A fundamental difference between the two models is that the IntServ model includes the definition of a signaling mechanism and an admission control framework. …