Multicast routing and recovery based in data center networks

Abstract

As a key part of supporting cloud computing, the performance of data center network directly determines the quality of cloud computing services. Multicast benefits data center group communication in both saving network traffic and improving application throughput. However, given data centers are experiencing an exponential increase in the amount of network traffic, multicast routing have to face the challenge of overload. In this paper, we present two algorithms: BLD (Based on Load and Distance multicast tree building), which is a novel multicast tree building algorithm for data center networks and RLPR (Reliable Lowest Prior Recover algorithm), which is an effective packet repair algorithm. The key idea of BLD is to build a multicast tree based on the load of server and distance between source and destination. This approach leverages the combination of ant colony algorithm and genetic algorithm to find the best solution, which effectively reduces the latency and improve the throughput. In RLPR, the packet loss rate of a node is estimated according to node's position, nodes with high packet loss rate are given priority to select their backup nodes. Finally, we verify the correctness and validity of algorithms through several simulations.