db-derby-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Henry <paul.he...@raytion.com>
Subject Insert/Update Performance
Date Fri, 18 Jul 2008 12:52:51 GMT

Hi All,

We have been using an embedded derby data base within our application for
some time now. However Recently we have added a new customer that has a
significantly larger amount of data within the application than existing
customers. The customer has been complaining about performance. We have
tracked the issue to our persistance layer.

We are using Hibernate layered over Derby. The database is relatively
simple, just six or seven tables, and we have added indexes where ever
required. So the Query speed doesnt seem to be a problem. However when
saving a new item into the database (An item will create 1 or 2 entries in
most of the tables and around 30 entries in 2 of the tables) we have been
able to measure a decay in the insert time as the db grows in size.

1) With an empty db we had measured delays of 31-140 ms to write a new item
into the DB

2) When the DB contained 1 million items (so a couple of the tables have ~30
million rows) we measured delays of 1000-1500 ms to store a new item

3) When the DB contained 2 million items we measured delays of 1600-2000 ms
to store a new item

Similar delays were recorded when performing an update of existing data.

Has anyone had any experience tuning the combination of hibernate over Derby
for databases with a similar table size, or does anyone have any advice how
to improve performance, whilst retaining both hibernate and derby?
-- 
View this message in context: http://www.nabble.com/Insert-Update-Performance-tp18528545p18528545.html
Sent from the Apache Derby Users mailing list archive at Nabble.com.


Mime
View raw message