xpath optimization

[Rob Redmon <Ro...@noaa.gov>]

Hi,

I'm wondering if anyone has any good performance/tuning notes. 
My scenario is thus:  I have need to build out a system that will manage 
on order of 2000 docs spread across 50 or so different collection paths 
and each doc having a nominal size of 30kbytes (as text file).  
Fragments of each doc will be exposed to a JSP front end mimicing a 
simple search engine.  The docs will probably all adhere to the same schema.

Maybe someone has some advice/notes/links/books they've liked on 
anything related, including topics like xpath, indexing, xsl, Java best 
practices, etc...  I'm new to xpath, so apologies in advance for any 
iodiotic questions.

Maybe this isn't a question for here, I guess maybe it's more of 
something I need to learn about xpath or the xmldb API.
I've been running into OutOfMemory exceptions when the result of an 
xpath query (via the Java API) is going to be fairly large (probably the 
result would be a few MB if stored as a xml text file) and so I'm 
wondering if there is some better tuning that I can do or better useage 
of XpathQueryService...

Thanks a bunch for any thoughts!

Rob

Re: xpath optimization

[Vadim Gritsenko <va...@reverycodes.com>]

Rob Redmon wrote:

> Maybe this isn't a question for here, I guess maybe it's more of 
> something I need to learn about xpath or the xmldb API.
> I've been running into OutOfMemory exceptions when the result of an 
> xpath query (via the Java API) is going to be fairly large (probably 
> the result would be a few MB if stored as a xml text file) and so I'm 
> wondering if there is some better tuning that I can do or better 
> useage of XpathQueryService...


First of all, avoid large files (see FAQ). If you can't live without 
large files (can't refactor into small files), please keep in mind that 
Xindice needs to have whole file in-memory in order to work on it 
(xupdate, xpath, etc). This means, that you have to give server enough 
memory (java options -Xmx, -Xms) to keep necessary amount of such 
documents in memory.

Vadim