We are currently on SQL Server 2005, with a plan to move to SQL 2012. To make this move we have to test new database drivers for the application (Java). We performed 2 load tests using the 2 different drivers.
- Transactions/sec = 9400
- Batch Requests/sec = 2400
- Transactions/sec = 7200
- Batch Requests/Sec = 6000
What I am interested in is why the massive increased in batch request/sec in test2, even though there was an overall drop in trans/sec (note that the front end performance mirrored the drop in trans/sec between the 2 tests). If it wasn't for us monitoring front end performance and trans/sec we would have actually thought we'd seen a 3x improvement in performance, instead of an actual 20% drop in performance.
Any thoughts on what features of the drivers may cause this?
For reference test 1 was performed with I-net Java driver, test 2 was performed with MS JDBC driver.