面向云计算的分布式应用自动部署框架-《计算机技术与发展》

文章信息/Info

Title:: Deployment Framework for Distributed Applications in Cloud Computing

作者:: 李超1 ; 花磊2 ; 宋云奎2; 1. 新华通讯社中国经济信息社有限公司,北京 100083;
2. 中国科学院软件研究所,北京 100190

Author(s):: LI Chao 1 ; HUA Lei 2 ; SONG Yun-kui 2; 1. China Economic Information Service Co. ,Ltd. ,Xinhua News Agency,Beijing 100083,China;
2. Institute of Software,Chinese Academy of Sciences,Beijing 100190,China

Keywords:: distributed application; application deployment; deployment framework; cloud computing

摘要:: 云计算环境下的分布式应用规模巨大、交互复杂、层次众多,增加了应用部署的复杂度。同时,应用通常同时部署在多个云平台上,而异构云服务 API 缺少统一标准,导致了多云之间不可互操作。针对以上问题,面向云计算环境,提出一种细粒度弹性应用自动部署框架 CADep,以管理云计算平台内部服务与基于云平台的业务组件之间的交互关系,并连接各种部署进程和资源。该框架使用轻量级的微内核实现了应用部署的核心机制,并通过可定制组件扩展部署功能。同时,通过分层描述应用及其执行环境,实现了细粒度的资源管理。实验结果表明,与当前分布式应用部署框架相比,CADep 具有较短的部署时间,有效减轻了部署的工作量,降低了部署的难度。

Abstract:: Distributed applications in cloud computing have many components,complex interactions and various layers,so deploying these applications is difficult. Furthermore,multiple cloud computing platforms have different service APIs,so these cloud services cannot cooperate with each other. For this,we propose a fine-grained flexible deployment framework CADep for distributed applications in cloud computing to manage cloud services and interactions with application components,and then integrate various deployment processes and
resources. CADep with a light-weight micro-kernel uses a component-based framework to improve the extensibility of application deployment. Furthermore,it uses the hierarchy description of applications and contexts to manage components in a fine granularity. Experiment shows that compared with typical deployment frameworks,CADep has short deployment time,reduces work amount and decreases operational difficulty.