Lucene intermittently takes longer as
1. It flushes the buffered docs from memory to the disk and
2. It merges the smaller index segments to form a larger segment on regular
intervals as per the index writer settings.
You may have a look at various IndexWriter params in the javadoc on the
lucene page, starting at
http://lucene.apache.org/java/2_9_1/api/all/org/apache/lucene/index/IndexWriter.html#DEFAULT_RAM_BUFFER_SIZE_MB
<http://lucene.apache.org/java/2_9_1/api/all/org/apache/lucene/index/IndexWriter.html#DEFAULT_RAM_BUFFER_SIZE_MB>
--
Anshum Gupta
http://ai-cafe.blogspot.com
On Wed, Dec 29, 2010 at 2:21 PM, King JKing <beuk...@gmail.com> wrote:
> Dear all,
>
> I use Lucene to index document content 6 field (int) and 1 file (string)
>
> I log the index process. Log said that,
>
> INFO [CONTENT-FILTER INDEX-TIMER] 2010-12-29 15:45:52,707 Index 55 item in
> 10576 miliseconds
> INFO [CONTENT-FILTER INDEX-TIMER] 2010-12-29 15:46:13,378 Index 19 item in
> 20670 miliseconds
> INFO [CONTENT-FILTER INDEX-TIMER] 2010-12-29 15:46:28,391 Index 35 item in
> 15013 miliseconds
>
> Why Lucene take much time for just index small amount of document?
>
> Could you give me the best performance of Lucence and How to improve
> performance?
>
> Here is my code
> Document doc = new Document();
> doc.add(new Field("system_id", data.systemId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("app_id", data.appId + "",
> Field.Store.YES,
> Index.NOT_ANALYZED));
> doc.add(new Field("object_id", data.objectId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("content_id", data.contentId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("owner_id", data.ownerId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("to_user_id", data.toUserId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("parent_id", data.parentId + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> doc.add(new Field("content", data.content + "",
> Field.Store.YES, Index.ANALYZED));
> doc.add(new Field("id", System.nanoTime() + "",
> Field.Store.YES, Index.NOT_ANALYZED));
> iWriter.addDocument(doc);
>
> Thank a lot for suppot.
>
