Cloud computing utilizes heterogeneous resources that are located in various datacenters to provide an efficient performance on a pay-per-use basis. However, existing mechanisms, frameworks, and techniques for management of resources are inadequate to manage these applications, environments, and the behavior of resources. There is a requirement of a Quality of Service (QoS) based autonomic resource management technique to execute workloads and deliver cost-efficient and reliable cloud services automatically. In this paper, we present an intelligent and autonomic resource management technique named RADAR. RADAR focuses on two properties of self-management: firstly, self-healing that handles unexpected failures and, secondly, self-configuration of resources and applications. The performance of RADAR is evaluated in the cloud simulation environment and the experimental results show that RADAR delivers better outcomes in terms of execution cost, resource contention, execution time, and SLA violation while it delivers reliable services.
KEYWORDScloud computing, quality of service, resource provisioning, resource scheduling, self-configuring, self-healing, self-management, service level agreement
INTRODUCTIONCloud computing offers various services like Software as a Service (SaaS), Platform as a Service (PaaS), and Infrastructure as a Service (IaaS).However, providing dedicated cloud services that ensure various Quality of Service (QoS) requirements of a cloud user and avoid Service Level Agreement (SLA) violations is a difficult task. Based on the availability of cloud resources, dynamic services are provided without ensuring the required QoS. 1 To fulfill the QoS requirements of user applications, the cloud provider should change its ecosystem. 2 Self-management of cloud services is needed to provide required services and fulfill the QoS requirements of the user automatically.Autonomic management of resources manages the cloud service automatically as per the requirement of the environment, therefore maximizing resource utilization and cost-effectiveness while ensuring the maximum reliability and availability of the service. 3 Based on human guidance, a self-managed system keeps itself stable in uncertain situations and adapts rapidly to new environmental situations such as network, hardware, or software failures. 4 QoS based autonomic systems are inspired by biological systems, which can manage the challenges such as dynamism, uncertainty, and heterogeneity. IBM's autonomic model 3 based cloud computing system considers MAPE-k loop (Monitor, Analyze, Plan, and Execute) and its objective is to execute workloads within their budget and deadline by satisfying the QoS requirements of the cloud consumer. An autonomic system considers the following properties while managing cloud resources 1-3 :• Self-healing recognizes, analyzes, and recovers from the unexpected failures automatically.• Self-configuring adapts to the changes in the environment automatically.In this paper, we have developed a technique for self-configuRin...