Thanx for the info. I just implemented it (that was quite simple!) and it seems to work ok and the way we want.. (no integration with the actual RDBMS yet, but a coworker will handle that)..
Mvgr, Martin On Thu, 2004-03-18 at 20:19, [EMAIL PROTECTED] wrote: > I use lucene to index several (currently 8) large > (greater than 3 million records each) databases. > These databases have a table structure that is > optimized to support the graphical display of large > numbers of alignments between biologically-relevant > sequences. Unfortunately, that structure is quite > suboptimal for knowledge-discovery type queries > (e.g., searching for all the sequences that are > characterized as being involved in a particular > biological or molecular function). So I export the > data and construct more biologically-informative > Lucene documents. Typical indexes are 500-600Mb per > organism, and I have a ParallellMultiSearcher using > RemoteSearchables to spread the search load over > several servers. Search speed is quite acceptable > (as long as post search sorting of results isn't > required). > > So I would say using Lucene to replace some of the > RDMBS query functionality is quite feasible and > yields good performance. > > --David Goodstein > Joint Genome Institute --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
