Global Dispatching of Distributed Cluster Flow Based on Dijkstra and Multi-Strategy in Power Grid

Affected by the uneven distribution of business traffic in the network, the principle of “many-to-one” data collection limits the process of finding the shortest path, resulting in a low degree of load balance, a trend of reduced throughput and unstable output results after scheduling, which leads to poor global scheduling effect. Therefore, this paper proposes a global scheduling of distributed cluster traffic based on Dijkstra and multi-strategies. Combining Dijkstra algorithm with multi-strategy theory, this paper predicts the distributed cluster traffic of power grid, evaluates the size of the traffic to be scheduled by using the constructed data center network traffic model, designs the distributed cluster traffic scheduling queue and multi-strategy scheduling framework, and applies Dijkstra algorithm to solve the optimal directed edge weight to find the shortest path, thus completing the global scheduling of distributed cluster traffic of power grid. The experimental results show that the average load balance of this method is 0.35, and the trend of effective throughput reduction gradually slows down. The output results after scheduling show high stability and good global scheduling effect.