Research On Service Dynamic Replacement Mechanism Model Supporting QoS In Cloud Service Composition

Cloud computing, as a newly-emerging model of business computing, has obtained considerable development. Software as a service(SaaS) which can be delivered in a composite form is one of the most common delivery model of cloud service running on the virtual machines in cloud computing environment. Meanwhile, the increasing tendency of network service users to use cloud computing encourages cloud service vendors to supply large numbers of cloud services that have similar functional but nonfunctional(Quality of Service) features and provide them in a service pool. In the highly complex and dynamic cloud computing environment, Cloud service composition,selecting appropriate single cloud services from a service pool and combining them into complicated service, is a typical NP-hard problem. As the dynamic nature of system operational environment, users requirements and especial component itself which participated in the combined system, service component may become unavailable, that will cause the disruption of cloud application.To solve this problem, this paper presents dynamic service substitution mechanism model supporting Qo S for cloud service composition, that is, when one cloud service component resulting from maintenance or unexpectedly lose power is unavailable,composite system can choose compatible and similar single cloud service component with higher QoS to substitute it, in order to avoid the disruption of cloud application.The primary research work of this thesis includes:1. Presents two definitions of service replacement: consistency replacement and equivalence replacement. This paper uses service automaton to model replaceable cloud service components, and introduces the Operating Guidelines to depict service replacement, and the Operating Guidelines matching algorithm is presented.2. Presents the method to Qo S global optimal service replacement driven by credit.A credit-aware per-service method to service replacement is proposed in this paper,which introduces credit level into service replacement model and is considered as Mixed Integer Linear Programming(MILP) problem. Cloud service composite system at run time can constrain service replacement according to the SLAs negotiated with the user and the high-availability of cloud service components perceived by itself, and then dynamically bind the best of a set of services to satisfy the customer's Qo S requirements.3. Proposes cloud service dynamic replacement model supporting Qo S QDRCM and cloud service dynamic replacement algorithm. QDRCM is built based on IBM'sMAPE-K(Monitor, Analyze, Plan, Execute, and Knowledge) reference model, and cloud service dynamic replacement execution flow and cloud service dynamic replacement algorithm are given.4. The paper designs and implements a prototype system, that is travel cloud service system(TCSS), to verify the validity of the proposed cloud services dynamic replacement method. Based on QDRCM model architecture, the implementation framework of TCSS is drawn and the dynamic replacement module is analyzed in detail.