Research on mechanisms of efficiency and reliability of data distribution servers

Abstract

For the general scenario of data distribution from one to many, the paper indicates that Internet includes large number of NAT and firewall equipments. In order to distribute data to all clients, the paper proposes that the clients use Web Service to fetch data. Taking into account that using Web Service will consume much of server's resource, the system uses both TCP Socket (to push data) and Web Service (to pull data) to obtain high distribution efficiency. In addition, the use of asynchronous socket technology improves the efficiency. The paper proposes Reachability Test Algorithm, which enables the server reliably identify which clients can not be connected by the server. So these clients should use Web Service to fetch data. Furthermore, the system uses cluster servers, hot standby scheme and load balancers to avoid single point of failure and balance the servers' load. Finally, the paper analyzes the system's reliability and efficiency, and an example is given to prove the high efficiency of the system.