Condor-G: A Computation Management Agent for Multi-Institutional Grids
HPDC '01 Proceedings of the 10th IEEE International Symposium on High Performance Distributed Computing
The Globus Striped GridFTP Framework and Server
SC '05 Proceedings of the 2005 ACM/IEEE conference on Supercomputing
User-friendly and reliable grid computing based on imperfect middleware
Proceedings of the 2007 ACM/IEEE conference on Supercomputing
The XtreemFS architecture—a case for object-based file systems in Grids
Concurrency and Computation: Practice & Experience - Selection of Best Papers of the VLDB Data Management in Grids Workshop (VLDB DMG 2007)
Efficient access to many samall files in a filesystem for grid computing
GRID '07 Proceedings of the 8th IEEE/ACM International Conference on Grid Computing
Globus toolkit version 4: software for service-oriented systems
NPC'05 Proceedings of the 2005 IFIP international conference on Network and Parallel Computing
Hi-index | 0.00 |
The abundance of middleware to access grids and clouds and their often complex APIs hinders ease of programming and portability. The Open Grid Forum (OGF) has therefore initiated the development and standardization of SAGA: a Simple API for Grid Applications. SAGA provides a simple yet powerful API with high-level constructs that abstract from the details of the underlying infrastructure. In this paper we investigate the price that possibly comes with such an API. We discuss the effects on expressiveness and ease of programming, and analyze the performance overhead of three different SAGA implementations (written in Java, Python, and C++) on various middleware. We conclude that SAGA is a good pragmatic approach to make grids easily accessible. The API considerably improves usability and uniformity, but offers a compromise between expressiveness and runtime dependencies. The overall performance of the tested implementations is acceptable, but the strict API semantics require various runtime checks that occasionally cause significant overhead, depending on the underlying infrastructure.