So the drillthrough functionality works great, until
your results take up all of your memory. Many of our
customers require drillthrough, and more often than not
the result sets are far greater than the memory we have
available. Sound familiar to anyone? Any solutions
Since we use xmla, we made some quick and dirty
modifications to XMLA_SOAP.executeDrillQuery - mainly
so that we do not use up memory, in favor of smaller
buffers. We stream the raw results to a temp file, then
likewise stream those results to an xsl transformation,
finally to the end user. Benefit is that we can handle
very large drillthrough results, only limited by how
efficiently the back end sql store can return results.
The negative is architectural, not very pretty, and not
true to jpivot design. I would prefer to submit the
code as opposed to having to apply this patch everytime
I deploy, but could use some feedback.
Log in to post a comment.