Mapeamento entre Arquitecturas de Serviços Integrados e de Serviços Diferenciados para suporte de Qualidade de Serviço na Internet
Authors
Abstract
The current trends in the development of real-time Internet applications and the rapid growth of mobile systems, indicate that the future Internet architecture will have to support various applications with different Quality of Service (QoS) requirements,The research effort in the area of the Quality of Service (QoS) provisioning in the Internet has been carried out by the IETF (Internet Engineering Task Force) according to two main approaches: the Differentiated Services (DiffServ) model and the Integrated Services (IntServ) model.
The IntServ model provides individually QoS guarantees to each flow. For such, it needs to make resource reservation in network elements intervening in the communication. For resources reservation the Resource ReSerVation Protocol is used (RSVP). The need of maintenance of state information on the individual flows is usually pointed as the origin of the scalability problems of the IntServ model.
The DiffServ model embodies the second approach where the flows are aggregated in a few Classes of Service (CoS) according to their specific characteristics. Different levels of QoS are provided to these classes.
In order to combine the superior scalability of the DiffServ model with IntServ superior QoS support capabilities, the ISSLL (Integrated Services over Specific Link Layers) working group of the IETF proposed the interoperation between these two models. The defined approach combines the IntServ model features ' the capability to establish and maintain resources reservations through the network ' with the scalability provided by the DiffServ model. The IntServ model is applicable at the network edge, where the number of flows is small, while the DiffServ model is applicable in the network core to take advantage of its scalability. The boundary routers between these two domains are responsible for mapping the IntServ flows into the DiffServ classes.
The main goal of present thesis is to propose an architecture that supports the mapping of the IntServ flows into the DiffServ classes. The proposed architecture and mechanisms pretends to complement the traffic control functions of the DiffServ network by using a dynamic Admission Control mechanism that reflects the state of network. In the adopted strategy, the decision of mapping and admitting (Admission Control) a new flow at the ingress of the DiffServ region is based on the behavior of previous flows which going to the same IntServ network. This behavior is evaluated by of delay and losses suffered by the flows in the DiffServ region. The underlying idea is inspired in the congestion control mechanism used by TCP, applied to the admission control and mapping of IntServ flows into DiffServ classes.
Even though the proposed architecture and mechanisms are generic, this work emphasizes the mapping between the Controlled-Load service of the IntServ model and the Assured Forwarding PHB group of the DiffServ model.
The results obtained by simulation show that the mapping mechanisms, including the admission control and the bandwidth management, extend the functionality of the IntServ networks through the DiffServ networks. The positive effect of the resources reservation in the IntServ model and the protection of the QoS characteristics of CL flows in the presence of best-effort flows were also verified. The proposed mechanisms improve the use of the available resources for the AF classes and detect QoS degradation occurrences and once detected they allow the reestablishment of the QoS characteristics of the AF Classes.