Hi, I'm getting pestered by developers and management regarding the estimated performance of Lucene. I've gotten the index running reasonably fast (on a corpus of 8 million docs).
In particular, some developers had been bitten by global GC on some pure Java webservers, especially when they loaded in a small number of objects. Note this wasn't Lucene... However, they were all very worried when I said I had a Gig of stuff in memory (the RamDirectory).... Has anyone had any experience with running high-volume searches using Lucene.In particular has anyone had any experience (or lack thereof) of GC problems ? (platform and JVM versions would be interesting) Was anything specific done in the design to avoid GC issues (I don't know if there is in general). I know the RAMdir is a long-lived object, and so should be immune to Global GC's but apparently Java still has issues in the area of Scavenges vs Global GC. I'm going to be running experiments, and will let you know my finding. Cheers, Winton Winton Davies Lead Engineer, Overture (NSDQ: OVER) 1820 Gateway Drive, Suite 360 San Mateo, CA 94404 work: (650) 403-2259 cell: (650) 867-1598 http://www.overture.com/ -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
