Attaining per Ow Qos with Class-based Diierentiated Services Attaining per Ow Qos with Class-based Diierentiated Services

The Diierentiated Services (DiiServ or DS) framework takes an edge over IntServ 1 because it is scalable and lesser complex. On the other hand, the application level end-to-end quality of service, in DiiServ, may get compromised because: 1) network resources are not allocated at microoow level (a data stream pertaining to a single connection) but at aggregate level (collection of one or more microoows), 2) the DiiServ working group does not specify algorithms for PHBs but their output behaviours and 3) end-to-end quality is function of Service Level Agreements (SLAs) between the adjacent domains transited by the connection and a large diversity in SLAs is quite evident as each DS domain would have diierent service provision policies. We focus, in this paper, on the rst two issues. Our goal is to have DiiServ deployed with all its simplicity and still be able to provide application level end-to-end quality of service. For that, we study a PHB for AF classes. A PHB comprises a packet scheduler and a packet accept/discard algorithm. For packet scheduler, we use the Extended-VirtualClock (Ex-VC) 2 algorithm. Ex-VC performs delay-based service diierentiation among the aggregates while selecting a packet for service. The reasons for having delay-based deenition for service diierentiation are: it is adaptable to load per aggregate and it does not need to be microoow aware. Other deenitions like bandwidth and loss may also be used but the former requires microoow aware management and the latter lacks in simplicity. 2 For packet accept/discard algorithm, we use RED when all packets have the same drop precedence level and DI-RO (Deterministic for In-RED for Out) when packets are policed at the ingress DS node and packets violating the Service Level Agreement (SLA) are marked OUT. In DI-RO, IN packets are always accepted (except buuer overrow) whereas OUT packets are accepted probabilistically.