Pegasus: Planning for Execution in Grids
Ewa Deelman, James M. Blythe, Yolanda Gil, Carl Kesselman · 2002
Grid computing has made great progress in the last few years. The basic mechanisms for accessing remote resources have been developed as part of the Globus Toolkit and are now widely deployed and used. Among such mechanisms are: § Information services, which allow for the discovery and monitoring of resources. The information provided can be used to find the available resources and select the resources which are the most appropriate for the task. § Security services, which allow users and resources to mutually authenticate and allows the resources to authorize users based on local policies. § Resource management, which allows for the scheduling of jobs on particular resources. § Data management services, which enable users and applications to manage large, distributed and replicated data sets. Some of the available services deal with locating particular data sets, others with efficiently moving large amounts of data across wide area networks. With the use of the above mechanisms, one can manually find out about the resources and schedule the desired computations and data movements. However, this process is time consuming and can potentially be complex. As the result it is becoming increasingly necessary to develop higher level services which can automate the process and provide an adequate level of performance and reliability. 2 Issues in Mapping Workflows onto the Grid In general we can think of applications as being composed of application components. The process of application development (shown in Figure 1) can be described as follows. The