Performance Evaluations of Geospatial Web Services Composition and Invocation

Abstract

Geospatial data and analytical functions are essential to geospatial modeling. There are increasing interests in publishing both geospatial data and analytical functions as Web services and use them as the building blocks for domain specific geospatial modeling. While the advantages of using the Web services technologies have been well recognized and a number of prototype systems have been built to demonstrate the feasibility, very few performance evaluations have been reported in the previous studies. Compared with business data, geospatial data is rich in data types, large in data volumes and complex in semantics. On the other hand, the Web services technologies are known to have significant overheads with respects to deployment and invocation. The answers to how effective the Web services technologies can be, and, to what extent they are effective under the typical computation environments for geospatial modeling remain largely unknown. In this study we have set up an experimental system by deploying several geospatial Web services on top of popular commercial and open source spatial databases and geographical information systems (GIS). The Kepler scientific workflow system is used for geospatial Web services composition and invocation. We have conducted experiments to chain the geospatial Web services into a geospatial model under two data volume levels and two network settings. Our experiments show that the geospatial modeling using the Web services technologies remains effective in the wired LAN computation environment for data volume as large as 10000 points. However, the same data volume level incurs significant response lags under the wireless WAN computation environment. The experimental results may be used as a guideline for geospatial modeling using the Web services technologies when performances need to be taken into considerations.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.