Flow Data Load Shedding Mechanism

With the development of information technology, more and more applications like net-work monitorin, communication service, sensor networ need to deal with a great deal of data on line. The system will be latent deteriorated and the system can not provide the timely response because the data stream is unpredictable, brusty, high, continuous, and the available system resources are limited. Under these conditons,the system needs to do some load management policy to improve quality of service.First, the timing of load shedding is very important in load management which includes these steps:the timing of load shedding, the criterion of judgment on load shedding. In this paper, a data stream load shedding model called Delay-Based Shedding Policy(DSP) is pro-posed which not only considers the processing capacity of system,but also the delay of processing data. The new policy can satisfy some applications which have strict require-ments on real time.Second, this paper also considers a kind of synopsis & summary algorithm based on cluster which can not only improve the velocity of process,but also can reduce the loss of precision. So this algorithm can guarantee the precision of queries.Further, the normal process is restricted by the maximum processing speed of singal processor,but the parallel processing can increase the maximum processing speed by multi-processor connection. this paper does some research on the parallel measure in order to improve the velocity and efficiency of process.Finally, in order to validate the performance of the load mechanism in real application, this paper applyes the whole method on the online journaling management system, and using this real application to test and verify the efficiency and practicability of load management.In summary, The study of load mechanism including load shedding, synopsis & sum-mary and processing data streams provides the theory and application support for further study.