Group leader election under link-state routing

In this work, we place a long-established distributed computing problem in a new context. Speciically, the group leader election problem is studied \in-side the network," meaning that participants in the election process are network switches/routers, rather than hosts. In this context, an election protocol can take advantage of certain internal operations of the network, such as the underlying routing protocol, in order to meet stringent fault-tolerance criteria while minimizing network traac overhead. A generic and robust solution to the problem, called the Network Leader Election (NLE) protocol, is proposed for use in networks that are based on link-state routing (LSR). The NLE protocol is \generic," in that it is independent of the leader selection policy, enabling it to support a variety of network services. The protocol is robust, for it achieves leadership consensus in the presence of adverse events, such as leader failures and network partitioning. The correctness of the protocol is proved formally. A simulation study reveals that the NLE protocol incurs low overhead in handling leader failures and in group creation. In addition, it is shown how important network functions, in both the Internet and ATM networks, can beneet from the NLE protocol. These functions include, but are not limited to, routing domain leader election, address resolution, and multicast core management.