| IDP (Integrated Data-Service Platform) is an open, modular, standards-based platform. Downwards, it provides communication capability with ISMG (Internet Short Message Gateway), MMSC (Multimedia Message Service Center), BOSS (Business & Operation Support System), WAP (Wireless Application Protocol) Gateway and other functional entities. Upwards, it provides user authentication, SMS (Short Message Service)/MMS (Multimedia Message Service) sending and editor, user management and other public functions. The main structure and the capability are on the basis of meeting existing mobile data services, including MMS, SMS and WAP. Many kinds of services can be conducted based on IDP, such as Mobile Paper, Mobile Advertisement, MMS Marketing and etc.Several kinds of services have been in the business market successfully. With the increase of kind and number, services'demand for usability and scalability of IDP has become up and up. To deal with the increasing load of IDP, it is essential to implement load balance based on cluster.Load balance is implemented by a server cluster which consists of multiple servers in a symmetrical form. Each server has the same status and is able to provide service independently. Through the specific load balancing technology, the external request will be distributed to a certain server rationally according to the load on each server in the cluster. It aims at increasing the speed of access to data and improving the server's overall processing capacity. Meanwhile, it intends to support a large number of concurrent accesses to application and improve reliability, availability, maintainability. The ultimate aim is to speed up response time, thereby enhancing the user's experience. This paper focuses on providing clustering support to IDP on the basis of load balance and proposing reasonable load-balancing algorithm according to the features of IDP.The problem IDP is facing is how to deal with large amounts of data in time. Specifically, IDP needs to construct and deliver large number of message packets to accomplish a message delivery task, which is quite different from the traditional load balancing system. We have to take into account the large difference between the load balancing system that IDP needs and the traditional load balancing system and conduct the special design.Based on the researching about basic principle of load balancing system and the algorithm about load balancing system, in view of the concrete characteristic of IDP, the self-adaptive load balance algorithm based on task partition (SALBTP) is designed. And then the SALBTP algorithm is implemented in the laboratory system of IDP and the laboratory system is systematically tested and the performance is valued. The result proves that the provided SALBTP algorithm occupies low computer resources, and the system is reliability so that it is adapted to the special environment of IDP.The fundamental idea of SALBTP algorithm is as follows. The SALBTP algorithm is based on the basic principle of negative feedback. Decision is made according to the usable resources and the load status of the load server nodes. According to the various weights of all server nodes, the big message delivery task will be divided into several subtasks with different size. Then each subtask will be assigned to a specific server node that has the matchable weigh with the subtask's size to adapt the fluctuation of the load dynamically. The server resources of all the nodes are fully utilized and the immediate response to the task is ensured after the adapting of the SALBTP algorithm. |
PDF Full Text Request