A cost model for distributed coverage processing services.

ABSTRACTGeo-scientific imagery with its high volumes and complex evaluation algorithms seems to be a good candidate for distributed processing. Typically today, algorithms offered via geo Web services are static, hence their distribution can be done manually in a relatively straightforward manner. This is different when it comes to SQL-style query languages and their flexibility. The Open Geospatial Consortium (OGC) Web Coverage Processing Service (WCPS) standard provides such a query language on multi-dimensional raster data. We present our research on an efficient distribution of coverage queries in a cloud based on a set of criteria. The paper's main contribution is a cost model for distributed processing of coverages. This includes determining the cost of executing different coverage data processing operations on different servers based on operations semantics, together with the costs of data transfer, data I/O, data decoding and encoding, etc. Important distributed coverage processing metrics and a server's calibration procedure for obtaining the metrics' values, as well as the experimental evaluation of the cost model are presented in this paper.