摘要：
云应用是云计算技术在应用层的一种重要体现形式,通常由分布式异构组件构成,且组件相互依赖,配置参数众多.组件依赖导致配置参数间存在关联,使应用运行时弹性扩展难以确定组件实例的配置顺序并保证关联参数的一致性,导致应用扩展后的系统故障和服务不可用.针对这一问题,提出了一种基于模型的云应用动态自配置方法,实现运行时组件实例配置顺序的自动协调,保障配置参数的一致性,提高应用运行时弹性扩展的可靠性.首先提出一种部署配置模型STM(Service-based Topology Model),该模型采用声明式的方法刻画云应用的部署拓扑结构,并基于服务的方式描述组件信息,实现组件间关联关系的分离,涵盖运维部署、扩展、运行时服务状态的监测.然后基于该模型,提出了一个云应用动态自配置协议,其基于服务注册发现机制实现组件间强依赖关系的解耦,保证动态调整应用实例时组件配置变化的一致性,实现组件部署配置的并行化.基于上述方法实现了一个原型系统,通过对分布式应用BookStore-TPCW的部署配置和运行时弹性扩展来验证方法的有效性.%Cloud-based application is one of the most important paradigms of cloud computing at application layer.These applications are usually constituted of a set of distributed components.Due to the dependencies between the components,there are correlations between the configuration parameters,which makes it difficult to configure the components correctly and to ensure the parameter consistencies at runtime,making system failed or service unavailable.To address this issue,we proposed a model-based self-configuration method for cloud-based applications,which realizes the automatic choreography for component configuration,ensures configuration consistency and improves the reliability of runtime scale.Firstly,this method proposes a model to specify application topology in a declarative way.This model describes application components in form of services,and makes the separation of components and their relations.Then this method designes a self-configuration protocol to support component configurations at runtime.Based on service registration and service discovery,this protocol decouples the component dependencies and ensures the consistencies of changing the configuration values.We also implemented a prototype based on this method and evaluated the effectiveness of this method with a typical application BookStore-TPCW.
