The Design And Implementation Of Network Performance Optimization At High-Performance Video Retrieval Platform

With the popularity of video information on network, the traditional information retrieval technology based on key words cannot keep up this development. As a result, content-based video retrieval has become a research "hotspot" in recent years. Currently, most search engines generally use the technology of cluster load balance, which makes the entrance of system easy to become network's bottleneck while high concurrent request of video search happens on network. For the goal of improving network performance of video retrieval platform, this paper focuses on the strategy of data concurrent access and the strategy of data cache, which are the key modules affecting service performance seriously.This paper puts forward three controlling strategies of data concurrent access:the strategy of service quality as high priority; the strategy of concurrency number as high priority; and the adaptive strategy of balance. For the first strategy, the system allocates all the resources for the current task, which ensures the shortest response time for single task while concurrency is still low. For the second strategy, the system executes multi-task of retrieval simultaneously in the way of threaded concurrency. Although the executive performance of single task cannot be improved in this strategy, the whole operating efficiency can be increased in the condition of high concurrency. With the adaptive strategy of balance, system will switch between the first strategy and the second strategy above smoothly according to the number of concurrent accesses and the special needs of video retrieval, aiming at improve the adaptive capacity of system farthest.To improve the QoS of video retrieval, this paper designs two kind of technology of data cache. In the method of retrieval result's data cache, by the construction of rapid lookup table of video content feature data, the pressure of video feature retrieval will be alleviated and the whole system performance will be improved. While in the method of retrieval task's data cache, by the construction of retrieval task table, the repetitive process because of the same tasks will be avoided, and then the pressure to the system from the retrieval of hotspot video will be decreased largely.This paper establishes testing environments and finishes the simulative experiments, which verifies the efficiency to improve the service performance of the video retrieval platform resulting from the strategies above.