Do Your e-call

Some service providers enhance the best-effort approach by adding differentiated service capabilities, to optimize resources for traffic of varying bandwidth and criticality – real-time voice and video, business-critical application traffic and best-effort traffic, for example – running simultaneously on their networks. Differentiated service adds state information to each packet – allowing the router to identify different service flows and direct queuing and forwarding treatment appropriate to the service requirements. This enables routers to identify voice packets and mark them for higher priority treatment over less sensitive packets. The IETF created this technique, known as Differentiated Services (DiffServ), as part of its Differentiated Service Code Point (DSCP) specification and the associated QoS framework.

To implement a Differentiated Services approach, the network administrator defines a number of service classes that corresponds to the number of QoS-dependant services that will be transported within a network domain. Each service class is assigned a 32-bit code (value) and a corresponding per-hop treatment. This treatment, often referred to as the Forwarding Equivalency Class (FEC) policy, infers handling priority and queue depth, as well as the rate policing or drop policy. The typical router interface can support eight different queues. Each queue can support a provisioned service class policy or DSCP. At the ingress interface, the router classifies each packet for queuing based on its DSCP value (or any single or combination of layer 2 and/or layer 3 parameters). The router services its queues according to a provisioned algorithm, such as weighted round-robin (WRR) and strict-priority queue (SPQ).

Service providers commonly define two service classes to handle bearer (audio) and signaling traffic in their VoIP solutions. Bearer traffic benefits from a high priority FEC with a shallow depth queue and no Random Early Detection (RED) rate policing drop profile. If voice is deployed in a network transporting only voice and one class of data, the FEC may be specified as a SPQ. But in converged multiservice networks that also carry video, no service should receive SPQ FEC treatment, as it will cause unwanted service delays in non-SPQ services that require high priority treatment.

The differentiated services approach enables service providers to manage QoS schemes by assigning voice traffic priority. However, because this approach provides QoS awareness independently for each node during packet transport, it does not provide any fault tolerance when the service needs it most. Additionally, the differentiated services framework alone does not support any proactive traffic engineering design capabilities.

Source: Juniper Networks, Inc. White Paper