Adaptable service level agreement evaluation framework based on dynamic monitoring interval in cloud computing

Service level agreement (SLA) is a contract between service provider and consumer in relation to the quality of service (OoS) in cloud computing. SLA contains agreed attributes and value of service level objectives (SLO). Service provider should deliver services based on agreed quality in SLA. For example, OoS value of running service exceeds the agreed SLO, service provider ought to bear certain amount of penalty or compound fees for the SLA violation. SLA monitoring tool is unavoidable to assess the agreed SLAs at run time and detect any probable SLA violations. Both service provider and consumer need to monitor and assess OoS to ensure SLA validity. The cost value and benefit value of SLA monitoring systems is a concerned issue in cloud computing. The SLA monitoring systems need resources such as CPU, Memory, or execution storage. The amount of consumed resources by monitoring system is the cost of SLA evaluation. On the other hand, SLA monitoring systems make benefits by detecting SLA violations in a sense that service provider subsequently can adapt the infrastructure to prevent more . numbers of SLA violations and avoid penalty cost. The interval value of monitoring system has a direct impact to the cost and benefit values of monitoring system. Experiment results for CASViD and LoM2HiS frameworks have demonstrated that short measurement intervals negatively affect the overall system performance, whereas long measurement intervals cause heavy undetected SLA violations. Current monitoring systems have often constant interval value and individual monitoring process for all services and SLAs which it decreases the adaptability and cannot make a balance between cost value and benefit value of SLA monitoring systems. Subsequently, service providers face either high monitoring overhead or significant number of undetected SLA violations. This study proposes SLA monitoring framework to evaluate the agreed SLA and detect SLA violations in cloud computing. The proposed framework locates a trusted-party between service provider and service consumer to maintain and evaluate SLAs, and without any manipulation of permission from service provider, the monitoring engine validate the SLA independently for each SLA data collector thread. This study also proposed a dynamic monitoring interval (OMI) to make a balance between cost values and benefit values of SLA monitoring system in different environmental situations (ES) monitored at run time. The experimental testbed is established with one server and four virtual machines (VMs). The developed monitoring tools evaluate the predefined SLA of a selected experiment case and produced results demonstrating that the proposed OMI have the highest profit in normal ES by 23.6% growth compared to the best static interval applied to the other four VMs. Each constant interval gained a high profit in specific situation and had a low profit in other ESs whereas the proposed OMI adapted the interval value successfully and achieved the high profit in all ESs compared to the static intervals