PROJECT TITLE :

Redundancy Minimization and Cost Reduction for Workflows with Reliability Requirements in Cloud-Based Services

ABSTRACT:

When it comes to the execution of workflows using cloud-based services, reliability requirement assurance is an essential quality of service (QoS) component. The enough replication for redundancy minimization (ERRM) algorithm and the quantitative fault-tolerance with minimum execution cost + (QFEC+) algorithm are state-of-the-art algorithms that can reduce redundancy and cost, respectively, for a workflow that has a reliability requirement. In this work, we propose the redundancy minimization using RIR (R RIR) algorithm and define the reliability increment ratio (RIR), which is used throughout the work. In addition to this, we present the concept of the geometric mean and propose an algorithm for the reduction of costs by using the geometric mean called C GM, which is based on the minimization of redundancy. The results of the experiments show that the proposed R RIR and C GM algorithms are superior to the algorithms that are considered to be state-of-the-art: (1) Although R RIR and ERRM show the same results for redundancy, R RIR is proven to generate minimal redundancy, whereas ERRM cannot; (2) R RIR only requires a few seconds to achieve minimal redundancy for large-scale workflows, and it has much higher time efficiency than ERRM; and (3) C GM generates less cost than QFEC+ in a large portion of cases.