Performance Evaluation of Hybrid Queuing Algorithms for QoS Provision Based on DiffServ Architecture

Based on the Differentiated Services architecture (DiffServ), a network is implemented in a controlled environment for Quality of Service (QoS) tests. A traffic generator is used to generate and transmit test traffic that simulates services and applications. Each service is classified and marked in such a way that the network layer manages the functions of providing a specific QoS treatment, prioritizing services in real time, and allowing non-priority applications to access the network. Subsequently, the network is subjected to congestion situations with overload in different scenarios with QoS policies and the Ethernet Service Activation Methodology (eSAM) performance test is applied. The mechanism used to manage bandwidth is Hierarchical Token Bucket (HTB) and it is combined with queue-based algorithms such as Per Connection Queueing (PCQ), Stochastic Fairness Queueing (SFQ), Random Early Drop (RED), and First-In First-Out (FIFO). The results are contrasted with the performance objectives agreed by the Service Level Agreement (SLA), and finally the feasibility of DiffServ model combined with hybrid queuing algorithms HTB-based to guarantee QoS requirements at the IP level is discussed.