Hi Suman, I couldn't find a link but talking about approaches to load an index into memory would be : 1. Create a tmpfs partition and copy your index into the partition, open the index reader/searcher from the tmpfs. * You would have to handle the copying/management of indexes in this case. * It is volatile i.e. would get wiped off in case the machine restarts. * Lets you create a partition of any size and therefore gives you the flexibility to have an index of > 4G even in case of a 32 bit machine/JVM architecture. 2. Use RAMFs, which would mean that the JVM copies the opened index onto its memory(using and sharing the JVM's memory and address space) * Index management would reduce. * Index size would be limited to the maximum address space available to the JVM process (In case of a 32 bit machine, it'd be 4G) * Would get wiped off when JVM goes off (Not affecting the Disk copy in any manner though)
-- Anshum Gupta Naukri Labs! http://ai-cafe.blogspot.com The facts expressed here belong to everybody, the opinions to me. The distinction is yours to draw............ On Tue, Mar 23, 2010 at 3:51 PM, <suman.hol...@zapak.co.in> wrote: > Hello, > > > I am trying for optimizing the searching by putting indexes onto > memory. RAMDirectory is not option for me, as I am transferring indexes > built to slave system to use. > So if u could let me know that how to place my indexes onto memory(m > thinking of using mmap) .For this I wanna know how exactly the lucene > searches indexes. Does it put all of the indexes onto memory, when we open > index searcher instance and then does searching. > > Curious to know how the lucene searches actually work.Can u get me the > link. > > > Thanks and regards, > Suman > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org > For additional commands, e-mail: java-user-h...@lucene.apache.org > >
