Scalable Reservation-Based QoS Architecture (SRBQ)

Having its roots in the military ARPANET, conceived as a data transport network with a focus on resilience, the Internet supports only a best-effort service model, where all packets are treated the same way, therefore providing a single level of service. Now that the Internet is becoming a ubiquitous global communication infrastructure, new applications are emerging with more demanding and diversified requirements than data transport. Internet telephony, for example, has much stricter delay requirements than remote terminal, the most demanding of the original applications. The deployment of other service models providing better quality of service (QoS) is of great importance for the transport of these new applications. While many approaches to providing QoS on the Internet have been proposed, none has gained enough acceptance to be deployed across the Internet at large. One important reason for this is the difficulty in conciliating strict QoS guarantees with the required scalability and with good resource utilization. The realization that the scalability limitations of the existing per-flow reservation QoS architectures are not intrinsic to the per-flow approach itself, but rather to the employed algorithms, led to the development of the scalable reservation-based QoS (SRBQ) architecture (Prior et al., 2003a).