Research On Dynamic Priority Scheduling Strategy Based On Improved Ant Colony Algorithm

Due to the development of cloud computing,big data,and industrial Internet,more and more application services are deployed into the data center network.Some data-intensive applications require that the flow can be processed completely as possible as shorter time.To minimize flow completion time(FCT)in the data center,it is ideal to assign a corresponding priority to each flow in the data center.The fact that there are limited priority queues available for flow scheduling in existing data centers has serious implications for minimizing flow completion time.Therefore,this thesis proposes a priority division strategy considering the waiting time and a path planning algorithm based on the improved ant colony algorithm.The main ideas of these two aspects are as follows:1.Considering the waiting time for the prioritization of strategy,not only to consider the size of the current data stream,consider the waiting time of flow,the priority of short flow,at the same time send ever-flowing had waited long enough for the opportunity,avoid to cause the ever-flowing hunger,and the effects of convection the overall average completion time is very small.And the strategy does not require the existing data center of the large number of switches provide priority and queue,because the policy only requires two physical priority queue,and,as a high priority queue,the other is called the low priority queue,when in the lower priority queues need immediate processing flow,and high priority queue is not empty,will trigger the polling algorithm,because it involves two priority queue,avoid the bye time overhead.2.In this thesis,the ant colony algorithm is improved,so that the ant colony algorithm can be used in the path planning of flow in the data center.The idea of the improved algorithm is that at the initial moment,the ant is placed at the source node,the candidate node set of its next hop is determined,and then the specific next hop is determined according to the transition probability.The transfer probability is determined by considering the value of ? and ?.The final step is to select the optimal path from the set of all accessible paths and then route.Finally,experimental results show that the proposed algorithm performs well in flow completion time,throughput and packet reception rate.