Proactive Automated Dependable Resource Management in Cloud Environments
Anna Schwanengel, Gerald Kaefer, Claudia Linnhoff–Popien · 2013
Abstract—Cloud Computing comes along with easy and self-managed resource provisioning and releasing. However, booting and shutting down of instances still means dealing with laten-cies, administrative efforts and supplementary costs. Considering that, scaling of required resources needs to be well-scheduled, especially, as resources in Clouds are often highly and complexly dependent among each other. The challenge, thereby, is to manage Cloud services with less overhead while fulfilling negotiated SLAs. To automatically provide the exact amount of required instances, our approach enables the detection of resource dependencies and the automated scaling of them without the need for observing the utilization of every single instance. For that reason, we introduce a model addressing resource dependencies and a self-calibration process of the dependency graph used by a regulation method for the dynamic management of dependent resources. Keywords—Resource Management, Dependencies, Cloud. I.