一种面向服务的网格作业管理机制

开放网格服务体系结构(OGSA)的出现表明让网格资源以服务形式提供标准化的接口已成为趋势，然而目前的网格作业管理系统主要针对以程序形式提交的批处理类型作业，其管理的资源对象和调度目标主要面向科学计算，在基于服务的网格环境下远不能满足应用的需求：一方面，用户的使用模式从批处理为主转向带有交互性质的服务访问模式；另一方面，不同的应用对服务质量(QoS)有着不同级别的需求，为了解决这些问题，提出了一种面向服务的作业管理机制，它作为用户访问网格资源(服务)的代理，为用户提供透明的、与资源物理位置无关的并带有会话支持的作业服务接口。还引入了服务水平协议(SLA)的概念来表示用户需求的不同网格服务级别，作业管理系统则根据可定制的服务水平实现配置(SLAP)将sIA中规定的各项QoS特性映射到具体的作业管理行为，该作业管理机制已应用于织女星网格系统软件中，并能够为基于服务网格的应用提供灵活有效的支持。

A Service-Oriented Job Management Scheme for the Grid

The emergence of OGSA represents a consensus on the standardization of resource accessing interfaces as grid services. However, current job management systems are not geared toward the application requirements from a service grid. The accessing fashion in a service grid has shifted from traditional batch processing to interactive communication with services. Moreover, different levels of QoS guarantees are demanded by different categories of applications. In such context, a service-oriented job management scheme is presented, which serves as the service broker mediating invocations between users and grid services, providing a set of transparent and location-independent interfaces. Service Level Agreements (SLAs) are also leveraged to differentiate QoS requirements, and the corresponding actions taken by the job management system could be customized via a profile named SLAP. The proposed scheme has been incorporated into the Vega grid. Preliminary results demonstrate its flexibility and effectiveness.