Splitting documents for performance...

[Norbert Hartl <no...@knowledgetools.de>]

Hi there,

I have some performance questions. I read
the FAQ and the part which is titled
"My 5 megabyte file is crashed..." was very
helpful in one of my cases.

If you don't have such an easy case
are there any guidelines how to split a document
while preserving the structure of it.

If we have a document which is like

<cars>
   <color name="red">
      <car id="1">
      .
      .
      </car>
      .
   </color>
   <color name="yellow">
      <car id="10000">
      .
      .
      </car>
      .
   </color>
</cars>

and we want to query car by attribute id how
would I split the document to gain performance.?
(let's assume the cars has 50 color childnodes
and every color node has 10000 car childnodes)

I presume that in the FAQ example after splitting
the document the car nodes aren't longer
subnodes of the cars node. Is this right?

Is there any possibility to split the document but 
preserve the structure? (maybe via collections)

thanks in advance,

Norbert