Formal Modeling of Self-Adaptive Resource Scheduling in Cloud

Abstract

A self-adaptive resource provisioning on demand is a critical factor in cloud computing. The selection of accurate amount of resources at run time is not easy due to dynamic nature of requests. Therefore, a self-adaptive strategy of resources is required to deal with dynamic nature of requests based on run time change in workload. In this paper we proposed a Cloud-based Adaptive Resource Scheduling Strategy (CARSS) Framework that formally addresses these issues and is more expressive than traditional approaches. The decision making in CARSS is based on more than one factors. The MAPE-K based framework determines the state of the resources based on their current utilization. Timed-Arc Petri Net (TAPN) is used to model system formally and behaviour is expressed in TCTL, while TAPAAL model checker verifies the underline properties of the system.