In the Internet today, learnt prefixes are forwarded within autonomous systems (ASs) over internal Border Gateway Protocol (iBGP) sessions. Existing schemes for iBGP routing include the full-mesh (FM) solution, route reflection (RR) solution and confederation. Optimal prefix routing and route diversity are the main strength of the FM solution. However, it is rarely employed in a large networks due to its deficiency in aspects including scalability and large Routing Information Base (RIB) size requirement of routers. This is due to the fact that routers in this topology are required to peer with every other router within the AS. To combat these challenges, the RR scheme provides solution for scalability by decreasing the iBGP sessions requirement. Notwithstanding, the RR solution has its own challenges which includes reduced route diversity, introduction of divergence and forwarding anomalies. Also, the FM optimality may be lost since the Route Reflectors are responsible for reflecting the learnt prefixes to their corresponding clients based on its partial view of the network. The concept of Software Defined Networking (SDN) entails decoupling of the control plane from the forwarding plane such that the control plane is logically centralized benefiting from an overall knowledge of the network for decision making.
In this work, we propose a solution based on multicasting which employs relay nodes in the iBGP message dissemination. Our solution brings session management scalability and minimization of duplicate prefix announcement through elimination of peer sessions deemed unnecessary. SDN controller is employed to configure and coordinate the multicast tree.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/36871 |
| Date | January 2017 |
| Creators | Bassey, Ukemeobong Okon |
| Contributors | Nayak, Amiya |
| Publisher | Université d'Ottawa / University of Ottawa |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0031 seconds