Saving downloaded time series is slow

jirikadlec2[CodePlex]
DROUGHT ANALYZER issue: In Data sources, Local Data source option, I select the data service
http://hydrodata.cz/chmi-d/cuahsi_1_1.asmx

I run search (all keywords), select all sites (total 510 sites) and select quotDownload dataquot.

The download phase is fast (less than 5 minutes) but the quotsaving to databasequot phase takes more than 25 minutes for 5 years of daily precipitation and temperature at the 510 sites.

Probably, this is a performance issue in the data sites layer either in the WaterML file reading or in the quotsave time series to databasequot step.

Reproduce the reported workload: 510 sites with five years of daily precipitation and temperature data. Profile the download, WaterML reading, and save-to-database phases to locate the bottleneck; done means the saving phase is measurably improved without changing the downloaded data.

Written by the indexing model from the issue text.