Ethernet Circuit Failure Detection over Aggregated Trunks

Abstract

With the increase in data-intensive research in recent years, the Ethernet circuit, which is a high speed point-to-point connection, can be used for transmitting large amounts of data between sites. Customers use the trunk port to connect to the operator network. It allows multiple Ethernet circuits to share the same trunk port by using the trunk and results in the efficient utilization of the bandwidth of the port. It distinguishes each (VLAN) service on the basis of VLAN identifiers. When redundancy needs to be offered in the network using two trunk ports, detecting an individual Ethernet-circuit failure over the trunk and load balancing per-flow traffic between active trunks is not possible because the existing technique, namely link aggregation, has limitations. Link aggregation does not support per-VLAN failure detection and must only be setup between directly connected network elements. Hence, it cannot be used for end-to-end failure detection when intermediate network elements are involved. In this thesis, alternative Layer 2 technologies are identified for detecting per-Ethernet circuit failure over trunk and per-flow traffic load balancing. Both traditional networking-based as well as software-defined networking (SDN)-based approaches are investigated to solve the aforementioned problems, and the findings are summarized. An SDN-based design to solve both failure detection and load balancing problems is proposed. Furthermore, the proposed solution is validated using proof of concept (POC) implementation. Finally, the performance of the POC implementation is evaluated and the findings are summarized along with recommendations for future work. Our findings reveal that existing Layer 2 technologies lack support in successfully detecting per-Ethernet circuit failure over trunk and per-flow traffic load balancing between active trunks. However, an SDN-based approach can successfully be deployed to solve both the problems.