Research of self-adaptation load balancing algorithm on streaming server cluster

This paper analyzed the characteristic of streaming server cluster and consider network link channel and node resource state synthetically, put forward a kind of improved adaptive algorithm: through dynamic feedback of node state, load balancer call predict and make load increment that limited by node state and guarantee load balanced distribution among service nodes, besides algorithm guarantee node load close to the true great load but not exceed that one which may influence node normal running. The results of simulation shows: there is certain improving in practicability, self-adaptation and load balancing in this algorithm.