A Data Stream View of Scientific Workflow - Grid Computing: Infrastructure, Service, and Applications - page 242

Schema
generator

Schema
optimizer

Grid

Runtime
optimizer

Scientific workflow

FIGURE 9.15 Execution of scientii c workl ow.

9.5
In this chapter, we introduced the current technique: a database view of
the scientii c workl ow. We have analyzed the scientii c workl ow from the
point of view of a data stream. The advantages and the constraints were
also given. From data model level, we explained why we chose the data
stream view for the scientii c workl ow. We then compared similarities
and differences between scientii c workl ow schemas and data streams.
Based on XML technology, we proposed a new way to process scientii c
workl ow based on automata-based i lters. Finally, with inspiration from
the XML data stream, we introduced the idea of using optimization
techniques over XML data streams for scientii c workl ow scheduling. In
future, we will further investigate data-stream-based techniques for scien-
tii c workl ow execution.

Conclusions and Future Work

References

1. SwinDeW-G Team. System Architecture of SwinDeW-G , 2007. Available at:
http://www.swinl ow.org/swindew/grid/, accessed on December 1, 2007.