The Design And Implementation Of A Load Balancing Scheduling System

CDN stands for Content Delivery Network. In CDN, we add a new level of architecture to existing Internet, placing the site's content in the nearest network node from the users. In this way users can improve the speed of response and reduce transmission pressure to the backbone network due to the large number of remote access. Users visit a Web site, by visiting the URL or IP address to connect to a access server, then access server redirect this user to another server through the visitor's geographic location, owned ISP and server deployment and load, etc.. Which server to be redirect needs to follow some principles, which is the nearest service and load balancing. In the actual CDN load balancing system, the access server use a load balancing strategy data to scheduling which can be produced on the basis of some statistical data.Load balancing strategy data can be seen as a mapping table which is from the "Area & ISP" to a list of server address, containing several resources server IP address, each with a probability. When an access comes, the access server queries the region and the ISP which the access is from, and then queries the load balancing strategy data. Then, a server address can be selected from the list of server address on the basis of probability which will be used to give service to the user access. Traditional CDN load balancing system use load balancing strategy data which was filled out by hand. The load balancing strategy data Filled out by hand can meet the basic needs of system, but has many shortcomings; such as scheduling efficiency is not optimal and when a server is idle the data need to be fill out again. Load balancing strategy data generated by the program data can circumvent these shortcomings. The load balancing scheduling system will be described consists of two parts, data generation and scheduling, the part of data generation use network flow algorithm to generate a load balancing strategy data, and the part of scheduling use this load balancing strategy data to scheduling. Load balancing strategy data will determine the quality of scheduling effects. Course, apart from the part of load balancing strategy data generation and scheduling, but also need other auxiliary parts, such as the number of visits Statistics and user management section.