Popis: |
One of the main goals in any computational system like the cloud is to effectively allocate resources proficiently for task scheduling. However, the dynamic characteristics of the cloud make it more prone to faults and failures. The flexible and responsive changes are made to redistribute virtual machines (VMs) to address these faults and failures for maintaining continuous services. However, it may inadvertently lead to uneven load distribution. Therefore, thorough attention is required to ensure carefully monitored load equilibrium following fault tolerance. Addressing all these issues simultaneously with optimized Quality of Service (QoS) parameters is a good need of time. In this paper, a novel hybrid model: the Hybrid Fault-tolerant Scheduling and Load balancing Model (HFSLM) has been proposed to optimize the makespan of the dynamically arriving tasks and efficiently utilize the available VMs. Moreover, the model also provides solutions for several crucial concerns in cloud systems including VM failure, and VM/task heterogeneity. In the consequence of a VM failure, the approach offers a Neighbouring VM as a substitute for the corresponding task to complete its execution. Furthermore, the model is escorted by a load-balancing algorithm to maintain the equilibrium of load distribution after fault handling for further optimization of the considered QoS parameters. HFSLM is evaluated by comparing it with FTHRM, MAX-MIN, MIN-MIN, and OLB on a small task scale over diverse task and machine heterogeneities and with ELISA and MELISA on an extremely large task scale. The evaluation results show that the recommended HFSLM tops the compared approaches in all the considered cases and heterogeneities. |