[KinoSearch] Search in a clustered environment

[Miles Crawford]

Hello -

I was hoping to get some advice on doing searching in a clustered environment.

We have four mod_perl tools servers that will be doing both reading and 
writing to a search index.  Right now, there are about 200 to 2,000 searches 
on a given day.

We have a shared NFS mount to all the cluster members, and I've seen someone 
talking here about just using an index stored on an NFS mount.  How safe if 
this on the 0.15 codebase?

I've noticed that the .20 release is coming, as they say, soon. ;)  Is there a 
firm estimate for purposes of planning dependent development?  I really like 
the commit-free reads, and feel like that will avoid much of the headache I'm 
asking for with NFS.

Does anyone have other suggestions for clustered searching that might be 
investigating?

Thanks,
-Miles

__________________________________
Miles Crawford, Software Developer
Catalyst Research & Development
Office of Learning Technologies
University of Washington
206.616.3406

http://catalyst.washington.edu
http://solstice.eplt.washington.edu