Jukka Zitting a écrit :
Hi,
On Thu, Oct 9, 2008 at 9:26 AM, Cédric Chantepie <[EMAIL PROTECTED]> wrote:
I would guess that most of the load you are seeing is used by full
text indexing. Especially the PDF parser we use to extract text for
indexing can be notoriously slow with large documents.
If you don't need full text searching, you can try removing some of
the text extraction classes (especially for PDF). See the
textFilterClasses parameter in workspace.xml.
Thanks I will try that.
By another way, maybe it's possible to make indexing more asynchronous.
Other than that I don't see any reasons why your setup would not be
production-ready.
PS. Note that enabling the data store will even further increase
performance (as fewer bytes are being copied around) in your case, but
I think you should be seeing pretty decent performance even without
the data store.
I'm testing with configuration which get best benchmark on our test server :
* LocalFileSystem
* PostgreSQLPersistenceManager (externalBLOBs, cacheSize=48,
minBlobSize=32768)
* DbDataStore (over PostgreSQL)
BR,
Jukka Zitting
--
Cédric Chantepie - mailto:[EMAIL PROTECTED]
Architecte des systèmes d'information
NOZICAA (Groupe SIGIRE) - http://www.nozicaa.com
20 rue de Sardaigne - Z.A. du Danemark - 72100 LE MANS
Tel: +33 (0) 243 82 97 97
Fax: +33 (0) 243 82 97 99
