Recently, we have made some changes to some views of SQL Server that could adversely affect performance. We decided to conduct some performance tests on these views to see how we influenced them. The results were unexpected.
The graph below shows the results of the performance tests that we performed. Here is what the graph represents:
- The blue line is the view before any changes are made.
- The red line displays the view after making the changes.
- The x axis represents iterations in the loop.
- Each iteration, a thousand new records are inserted (which will return a view).
- At each iteration, we make several samples from the form under consideration and average the results.
- The y axis represents the time it takes to present the results.
- The select statement, which has been tested by performance, has a where clause so that it only returns 100 records each time. (during the test, 100 entries were added for each name).
The results show us that we have definitely achieved success in productivity, but what confuses us is a huge increase in productivity after we collected about 40,000 records in the database. We conducted this test on several different servers and get the same results every time.
, - , . , 40 000 ? - - ? - , .
, , , , , .
. .