Hello,
I'm going to index a lot of data in lucene. I have just written a functional prototype. One requirement for the application I'm designing the prototype for are the ability to present the search results ordered by date and since the data is very frequently changed I can't have the IndexReader (for the searcher) open for more than 5 minutes. I'm hoping that I don't will have to redo the ranking functionality for this so I started by trying to sort the results using the document dates tick representation. The prototype is currently working on a relatively small index ( around 8 GB ) and I'm already having performance problems. If I query the index without sort the response times are around 100 ms if I add a sort to the query the response time are around 12 seconds (even for small resultsets). I can't figure out what's taking so long? Doesn't the retrieval of field cache only occur at the first sorted search as long as I maintain the same IndexReader (could this be a bug in the Lucene.Net port??) ? / Regards Marcus
