Fault tolerant virtual machine consolidation for energy-efficient cloud systems

The requirement of optimizing large hardware and reducing their costs while considering the demand of software infrastructure organizations contributes Cloud computing to get more popular. Besides, automatic scaling, virtualization of hardware and guaranteed service performance features of Cloud computing increase the ease of adoption.On the other hand, the rise of Cloud computing applications may cause enormous energy consumption. Therefore, while satisfying the Cloud computing features, Virtual Machines (VMs) may be consolidated based on their utilization in order to balance the load or optimize energy consumption of physical machines.Since, these physical resources are vulnerable to system failure or in the need of maintenance, the probability of unavailable physical resources and coordination of the VMs over physical machines should be considered by a proposed optimization model.Otherwise, it is possible to encounter adverse results such as various unnecessary application migrations and consolidations of VMs that may easily lead to Service Level Agreement (SLA) violations and inefficient use of energy. In this study, a new model is suggested to decrease energy consumption and decrease number of application migration without violating SLA, while considering the fault occurrence possibilities in the cloud environment. Simulation software is developed by using Java 8 for this study. The main flow of the simulation could be separated as two main parts; application assignment to VM and VM consolidation on a physical machine.In the first part of the simulation, the applications are placed to VMs by using Round Robin algorithm. After that, the second part of the simulation checks all the VMs for consolidation; according to the selected algorithm. In the simulation, there are three types of algorithms could be selected; no-consolidation algorithm, consolidation by threshold algorithm and Fault Tolerant Consolidation (FTC) algorithm. Comparison of energy consumption on different threshold values, changes of energy consumption and number of application migrations during the simulation, by using different Guaranteed VM (GVM) selection policies and comparing application lifetime in the case of cold migration there are four different experiments has done to prove the performance of the FTC algorithm.Experiment results show that proposed approach decreases energy consumption by approximately 37% and the number of application migration by approximately 9%. Furthermore, when the fault occurs, the rise of energy consumption is less than 11% with the help of proposed model.