Research On Dynamic Resource Supply Method In Cloud Environment

In recent years, cloud computing has achieved rapidly development and gradually becomes a hot issue in both academia and industry with wide attention and discussion. A feature of cloud computing is that it delivers services that are made available to consumers in a pay-as-you-go model. However, in order to deliver IT services according to predefined SLA or QoS goals and minimize SLA violations, service providers always keep them in high performance. On the other hand, variations in workload demand make it difficult to provisioning resource appropriately. Static provisioning cannot fit all, and will result in either over-provisioning or under-provisioning. Hence, proper resource provisioning while minimizing data center SLA violations has become a challenge under cloud environment.From the perspective of an agile resource provisioning, we present the design, implementation and evaluation of an approach which combines predict-based proactive provisioning and reactive provisioning. Our proposed dynamic resource provisioning approach analyses the workload features, predicts the future demand and adjust the capacity of the underlying resource. Workload analysis generates the information about periodic patterns and ARIMA-based predictor gives trend of the demand. This is the proactive provisioning in the long term. We use reactive provisioning to handle estimation errors in the predictions or abrupt situations.At last, we design a prototype to validate our novel dynamic resource provisioning approach and carry simulations in a large scale. The experiment results show that our approach achieves high predict accuracy and has a proper resource provisioning while ensuring SLA and reducing power consumption.