Research Of Fault-tolerant And Load-balance Technology Of Redundant Network In Distributed System

Network redundancy mechanism has been widely applied in distributed systems which require high-availability among nodes in network communications. High availability is usually embodied in the following three levels: system level,component level and link level. Redundant deployment makes the physical links among nodes backup each other. All links share the load evenly to promote the network performance as they work well, while one or more links failed, system can use the remainder to ensure the network communication work well.At present, there exists some technologies to implement the fault- tolerance and load-balance of network such as SUN Trunking,Cisco Etherchannel,Linux bonding and so on. Usually, these technologies, also named as link aggregation or multi-NIC bonding, must base on special hardware platform or tend to result in single-point malfunction. For these reasons, it is difficult to implement peer-to-peer communication among nodes in distributed system.In this topic, we have studied the principle of reliability in network communications, development of network redundancy mechanism and its engineering applications. Unlike the technologies discussed above, the technology implemented in this paper virtualizes the physical redundant networks as single logic network and has not changed the deployment and working manners of current physical network. Its purpose is designing a universal redundancy solution which holds high-reliability for network. This solution enables the local node to communicate transparently with other nodes in system via a single virtual network and makes the designer not consider the lower layer communication details.This thesis is divided into six chapters. Chapter 1 discusses the background and significance of the topic; Chapter 2 has introduced briefly the existing technologies at present; Chapter 3 has introduced the related technologies used by this topic; Chapter 4 has discussed the principle of fault-tolerance and load-balance implemented in this topic; Chapter 5 has discussed the algorithms used by the implementation; Chapter 6 is mainly a summary of the whole thesis, and also has pointed out the following job to improve the design.