I have about 1000 xml files loaded, that average about 200kb each.
I (tried) to build indexes on all elements and attributes.
In my application, wildcard-browsing queries are important.

I.E. Users will want to explore unfamiliar data sets with queries like:
//[EMAIL PROTECTED]'SLIDE_NUMBER']

With my dataset (using the lastest CVS version) such queries
take from 20-50 seconds to complete.

I loaded the same dataset into a 30-day trial vesion of eXcelon.
The same query, on eXcelon, completes in 1.4 seconds.

Have I built the Xindice indexes improperly?
Or is that the way it is  right  now?

$XINDICE_HOME/bin/xindiceadmin add_collection -c $SERVER:$PORT/db/mycollection -n data

$XINDICE_HOME/bin/xindiceadmin add_indexer -c $SERVER:$PORT/db/mycollection/data -n starE -p "*"
$XINDICE_HOME/bin/xindiceadmin add_indexer -c $SERVER:$PORT/db/mycollection/data -n starA -p "[EMAIL PROTECTED]"


$XINDICE_HOME/bin/xindiceadmin import -c $SERVER:$PORT/db/mycollection -n data -f data

sample data:
<?xml version="1.0"?>
<XSIL Name="user_description">
<XSIL Name="system">
  <XSIL Name="displayView" category="CCB">
    <XSIL Name="displayView" category="BBT">
      <XSIL Name="displayView" category="saddam" />
    </XSIL>
  </XSIL>
  <Param Name="description">Afferent_bbt_L.10.9.538</Param>
  <Param Name="owner">saddam</Param>
  <Param Name="group">CCB</Param>
  <Param Name="template">BBT</Param>
  <Param Name="doc_id">BBT-saddam-01_01_1994-23</Param>
  <Param Name="creationDate">01_01_1994</Param>
</XSIL>
etc.....

--
/* Sandy Pittendrigh  >--oO0>
* [EMAIL PROTECTED]
* http://cns.montana.edu/~sandy */




Reply via email to