Mailing-List: contact java-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: java-user@lucene.apache.org
Received-SPF: pass (asf.osuosl.org: local policy)
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C6A10D.FA272DDA"
Subject: Berkeley DB JEDirectory Performance
Date: Thu, 6 Jul 2006 17:06:45 +0200
Message-ID: <D4E368C6FE4C57499CBCE2A0B476C8B05CE777@nebukadnezar.insiders.de>
Thread-Topic: Berkeley DB JEDirectory Performance
thread-index: AcahDcusnJv7SFOSTJawuNELqpBwRQ==
From: "Johannes Christen" <j.christen@insiders.de>
To: <java-user@lucene.apache.org>

------_=_NextPart_001_01C6A10D.FA272DDA
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Hi all.
=20
I just want to share my experience with the Berkeley DB JEDirectory
implementation from the contrib. area. I spend two days evaluating and
testing it and found out that it does work, but has very bad performance
and very high disk requirements for medium size document volume.=20
=20
I indexed about 78000 documents (DPA news items) in the FSDirectory and
the JEDirectory, and here are the results:
=20
Disk usage (index size):
FSDirectory: 322 MB
JEDirectory: 4650 MB
=20
Indexing Performance:
FSDirectory: 84 minutes
JEDirectory: 402 minutes
=20
Searching:
Initial opening of the JEDirectory took about 45 minutes.
The searching itself was ok, but still about 1.5 times slower than with
the FSDirectory.
=20
Ok. I hope than helped people who consider using the Berkeley DB
directory implementation in their application. It may do a good job if
you want to use transactions in small environments, but if the amount of
documents is getting big I wouldn't recommend the JEDirectory
implementation.
=20
Bye for now
=20
    Jo Christen
=20
=20

------_=_NextPart_001_01C6A10D.FA272DDA--