Joe,

  Put the data to be returned in a single field, and ideally in a 
static final array in memory. Get Lucene to return a simple key (or 
better yet use the internal Doc ID). Retrieval should be fast. Also, 
make sure you aren't retrieving too many hits (the hit collector is 
differently optimized for diferent query types).

Network shouldn't be too much of an issue (just establish a base line 
load with a static page serve).

Cheers,
  Winton

At 13:42 -0500 2/1/02, Joe Lerner wrote:
>We built a Lucene prototype, and compared it to our current Search Engine,
>Verity's K2, running on Linux.  Query times are quick, but performance of
>the combined time to return hitlist data and transfer it across the network
>was slow.
>
>Any tuning suggestions?
>
>
>-Joe
>
>
>
>--
>To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
>For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>


-- 

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]>

Reply via email to