3D Analytical Modeling and Iterative Solution for High Performance Computing Clusters

Mobile Cloud Computing enables the migration of services to edge Internet. Therefore, high performance computing clusters are widely deployed improve computational capabilities such environments. However, they prone failures and need analytical models predict their behavior in order deliver desired quality-of-service quality-of-experience mobile users. This article proposes a 3D model problem-solving approach for sustainability evaluation high-performance clusters. The proposed solution uses an iterative obtain measurements overcome state space explosion problem. availability modeling master nodes performed using multi-repairman approach. optimum number repairmen is also obtained get realistic results reduce overall cost. validated discrete event simulation. much faster good agreement with simulations. analysis focuses on mean queue length, throughput response time outputs. maximum differences between simulation considered scenarios up billion states less than 1.149, 3.82, 3.76 percent, respectively. These well within 5 percent confidence interval model.