 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Java Forum / Databases / August 2007Apache Derby performance disappointing
Hello, I am currently adding Derby support to an application that already has support for multiple DBMS via JDBC. I have found it to be significantly slower than SQL Server or MySQL when performing the same task. To provide this, I have created an exact duplicate database within both Derby and SQL Server and ran the same query on both, gathering timings. The test query returns about 16,000 rows from a table with about 100,000 on it, grouping data using a timestamp that is converted to the nearest second. This is one of the heaviest queries within the application and what chosen to put the DBMS under the most strain. To prevent caching interferring with the result, I made sure to restart the DBMS before running the query. SQL Server was at least 4 times quicker returning the results compared to Derby. I had expected SQL Server to outperform Derby, but not by such a huge factor. Is this result what is expected? The initial test was based on a default, untuned installation. I then followed tuning guidelines for Derby, increasing the page size to 32K and increasing the maximum page count to far in excess of what should be needed to return the amount of data I am testing. This gave some improvements but it is still takes three times that of the same query time on SQL Server. As Derby is marketing itself on the same level as MySQL (Apache provide performance figures comparing it to this) then I anticipated similiar performance, but instead it is taking twice as long as I consider reasonable. Is this in line with anyone elses experience? > Hello, > [quoted text clipped - 27 lines] > similiar performance, but instead it is taking twice as long as I > consider reasonable. Is this in line with anyone elses experience? I presume you created all indexes which exist on SQL server? The database schema and content are identical. All tables, views and contraints including indexes exist within both. In your experience, would you expect Derby to be so much less performant? >In your experience, >would you expect Derby to be so much >less performant? Sun typically puts out only reference implementations, not high performance ones. It does not want to put its customers out of business. If it did, it would discourage anyone developing Sun tools for fear Sun would use its money and expertise to destroy them.  SignatureRoedy Green Canadian Mind Products The Java Glossary http://mindprod.com >> In your experience, >> would you expect Derby to be so much [quoted text clipped - 4 lines] > developing Sun tools for fear Sun would use its money and > expertise to destroy them. And ? SUN Java DB is just a Derby with support. Derby is developed by Apache. They do not hold back on developing products to please commercial vendors. (before Apache it was IBM and Informix that owned the database under the name Cloudscape, but it has been under Apache for 3 years now) Arne > I am currently adding Derby support to an application that already has > support for multiple DBMS via JDBC. I have found it to be [quoted text clipped - 13 lines] > to Derby. I had expected SQL Server to outperform Derby, but not by > such a huge factor. Is this result what is expected? My experience show that Derby perform well in INSERT, but are clearly behind in SELECT. So it may very well be as it is. Arne > Hello, > [quoted text clipped - 4 lines] > within both Derby and SQL Server and ran the same query on both, > gathering timings. You really should consider subscribing to derby-user and ask your question there. To get help you probably need to include some more details about your schema and the load you are running. > As Derby is marketing itself on the same level as MySQL (Apache > provide performance figures comparing it to this) then I anticipated > similiar performance, but instead it is taking twice as long as I > consider reasonable. Is this in line with anyone elses experience? For a given type of load that is true. It may not be for your load. Signaturedt Free MagazinesGet these publications absolutely FREE for up to 12 months. There are no hidden fees and no obligation. Simply choose a title, complete the application form and submit it. Read more ...
|
Reply to this Message
|
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.