A Novel Approach for Energy Efficiency Load Balancing and Fault Tolerance in Cloud

Abstract

Cloud Computing is one of the fastest growing technologies today used for supporting a wide range of applications in diverse industries such as Information Technology, Healthcare, Transport, Entertainment etc. This is done by providing various services, the list of which is growing exponentially. Due to the enormous growth of cloud computing applications in recent years, this has resulted in the rise of numerous data centers, which can lead to massive amounts of energy consumption. The vast utilization of electricity has resulted in a variety of environmental issues due to the release of massive amounts of carbon and an increase in the maintenance costs of data centers. On the other hand, providing energy optimization, load balancing and fault tolerance in cloud is also an important field of research. In this work, firstly, we have presented a novel strategy, Dynamic Voltage and Frequency Scaling (DVFS) based Adaptive Cloud Resource ReConfigurability (ACRR) for distributed computing gadgets. This strategy reduces energy utilization as well as performs operations in very less time. Secondly, we have demonstrated a productive asset assignment and have used strategy for decreasing various expenses by the proposed ACRR model. We have also demonstrated effective energy-saving method by reducing errand loads. Our experimental results show that our proposed model ACRR outperforms other existing methods in terms of normal run time and power utilization. Further, we have designed an efficient mechanism named EAPT (Energy-Aware Parameter tuning) to minimize energy consumption. In here, three significant parameters such as computation cost, communication cost and task size are tuned for optimization, resulting in energy minimization by dynamically balancing the load. EAPT tunes this in efficient way such that loads are distributed based upon resources. EAPT is evaluated by considering the energy consumption as a parameter and varying the number of VM as 20, 40 and 60. Moreover, a comparative analysis is carried out with the existing DVFS model. Comparative analysis shows that EAPT consumes marginally less energy than the existing model. Simulation driven framework has been designed for improvising the fault tolerance. Furthermore, we have developed Redundancy Aware Cost Optimization (RACO) to achieve the reliability and to optimize the cost in workflow model. RACO mechanism is parted into the different sub-mechanism which includes the fault tolerance optimization, solving the reliability issue and later optimizing the cost. Lastly, RACO is evaluated by considering various scientific workflows and the cost parameter thereby, making our model outperform the existing model.