EA Heuristic Optimization Algorithm for Resource Load Balancing Based on Random Forest Model in Web Cluster

Abstract

Web cluster system needs to use all kinds of resources of the system to meet the customer's
request. How to make the system resource balanced distribution and make the optimal
utilization of system resources is an urgent problem. Stochastic forest model improves the
prediction accuracy of the model by summarizing a large number of classification trees. It is a
new model to replace the traditional machine learning methods such as neural network. Aiming
at the current application status and characteristics of Web cluster, a heuristic algorithm is
proposed, which can balance the load of resources and make full use of system resources. It is
an extension and optimization of set partitioning problem (SPP) and multiple choice multi
dimension knapsack problem (MMKP). The heuristic algorithm can significantly reduce the
computational complexity in the optimal allocation of resources, and make it meet the needs of
real-time scheduling. The simulation results show that the method is effective.