Distributed System Load Balancing Strategy for Converged Workstations

Abstract

In the era of cloud computing and big data, distributed block storage system has become more and more important with its unique advantages. Load balancing (LB) is an important feature of distributed block storage systems, and it is also one of the hotspots of current distributed block storage research. Based on the existing LB algorithms, researching new and more effective LB techniques is very effective in improving the stability of the system. The main purpose of this paper is to study the LB strategy in the distributed system (DS) of converged workstations. Based on the traditional LB strategy, this paper selects multiple evaluation indicators and establishes a corresponding evaluation function to improve the accuracy of judging the actual load of the node server. According to the situation of different hardware parts, the hierarchical standard of the load state is formulated, and the instantaneous stress test of the improved LB algorithm and the weighted minimum connection algorithm is carried out. The complexity of a LB algorithm. Experiments show that with the increase of the number of request connections, the response time of the system under the two algorithms is longer, but the response time of the improved LB scheduling algorithm has a smaller increase. That is to say, the improved LB algorithm is better than the weighted least connection algorithm in handling the instantaneous pressure of the system.