db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel John Debrunner <...@debrunners.com>
Subject Re: [jira] Commented: (DERBY-110) performance
Date Mon, 27 Dec 2004 17:28:50 GMT
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

zhangjinsheng wrote:

> Both hsqldb and derby are embedded, I haven't do any settings (all are in
> default).

With your test Derby and hsqlb are not performing the same
functionality, thus you have to consider if this is a reasonable
performance test.

Each INSERT statement is a seprate transaction due to auto commit mode,
auto commit is on by default as defined by the JDBC spec.

Derby is guaranteeing that if the system crashes in any way that any
committed transactions will remain committed, in this case each insert
statement has an implicit commit. This guarantee requires a forced write
to the disk for every commit, this is where Derby is spending most of
its time, waiting for the data to be flushed to disk.

I do not belive hsqldb is providing this guarantee, thus it goes faster
because it is not waiting for the disk after each statement.

This thread has some info on the differences between Derby and hsqldb.

http://nagoya.apache.org/eyebrowse/BrowseList?listName=derby-user@db.apache.org&from=630545&to=630545&count=10&by=subject&paged=false

Dan.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFB0EZSIv0S4qsbfuQRAi84AKC9z7zxB5PL2AuF5nUdxe1EFUrJdQCfdkjo
1MOlSub9SBYbsGCPZ1Q+pQU=
=GBZ5
-----END PGP SIGNATURE-----


Mime
View raw message