Difference between revisions of "Base/Performance"
|Line 69:||Line 69:|
Latest revision as of 20:40, 21 April 2009
|About this template|
This page contains all thought and ideas we have around performance.
At the moment, this is just a basket to hold items I needed to remove from other pages. On the medium term, this is to be the central entry point for performance considerations in Base.
- opening forms
- starting the Java-based wizards
- creating a new HSQLDB (which implies starting the JVM)
- saving a form
- flickering when starting the New Database wizard from within the start center
- startup time when selecting the Tables icon on the graphic interface after initial launch of OOo
- Import wizzard is reported to be very slow with large recordsets.
- Query designer does not know JOIN, so all queries use WHERE even if JOIN would be more effective.
- Using the "Binoculars" button in the GUI to find data.
We currently evaluate how row fetching can be accelerated. Fetching thousands of rows isn't as fast as it could be.
|Nr||Test Case||Source Database||Destination Database|
|1.||Jump to last row (65514 rows x 21 columns)||Flat (csv)||Flat (csv)|
|2.||Jump to last row (20000 rows x 11 columns)||MySQL (JDBC)||MySQL (JDBC)|
|3.||Open a spreadsheet file with one table (65514 rows x 21 columns)||Spreadsheet||Spreadsheet|
|4.||Jump to last row (65514 rows x 21 columns)||Spreadsheet||Spreadsheet|
|5.||Copy a table (13000 rows x 19 columns)||HsqlDB (native)||HsqlDB (native)|
|6.||Copy a table (13000 rows x 19 columns)||Spreadsheet||HsqlDB (native)|
All test results are normalized to 100% which is the worst case. First results which could be achieved so far (less is better):
- Test Case 1
- Test Case 2
- Test Case 3
- Test Case 4
- Test Case 5
- Test Case 6