XML Data Source Performance

jjfeigal

We are currently working with BIRT 2.5.2 and we are starting to write BIRT reports that access XML data sources. The reports look great, but for large data sets the report generation takes logner than we had hoped.

For example, for testing purposes we have generated an XML file that contains 300,000 "records". The file has the following structure:

<view>
<viewEntry>
<elementFirst>...</elementFirst>.....<elementLast>...</elementLast>
</viewEntry>
.
.
.
</view>

In our environment we have a data server, and a reports server. BIRT accesses the data across our internal network.

One of our reports will process the XML file using an XML data source in order to generate a report using a data cube. The an XML schema is provided for the XML data source so that only one pass through the data is needed. It takes 10 minutes and 20 seconds to generate the report. There is increased network activity for the first 7 minutes and 30 seconds of report execution time. I'm assuming that file access is taking place during this time and that the remaining time is associated with creation of the data cube and generation of the report.

Does anyone have any suggestions for improving report performance when processing large XML files?