出版社：SISSA, Scuola Internazionale Superiore di Studi Avanzati
摘要：The Grid approach provides a uniform access to a set of geographically distributed heterogeneous resources and services, enabling projects that would be impossible without massive computing power. Grid computing both provides the resources that allow scientists to run jobs over large amounts of data and also allows these data to be distributed all over the world, through the storage systems. Different storage projects have been developed and many protocols are being used to interact with them such as GsiFtp and RFIO. Moreover, during the last few years various Grid projects have developed different middleware such as gLite, CondorG, ARC, etc., and each one typically implements its own interface. For a user community that needs to work on the Grid, interoperability is a key concept. An efficient usage of the distributed resources is required to manage the heterogeneity of the involved resources. To handle different Grid interfaces and the resource heterogeneity in a really transparent way we have developed two modular tools: BossLite and Storage Element API. These deal respectively with different Grid schedulers and storage systems, by providing a uniform standard interface that hides the difference between the systems they interact with. BossLite transparently interacts with different Grids, working as a layer between an application and the middleware. Storage Element API implements and manages the operations that can be performed with different protocols used in the main Grid storage systems. Both the tools are already being used in production in the CMS computing software which uses the Grid services and resources for distributed analysis and Monte Carlo production. In this paper we describe their implementation, how they are used and give some result of performance measurements.
