Jimmy Nilsson
Rev: 2002-10-12
I ran some quick and dirty comparisons of the performance for the three methods of optimistic concurrency control that I discuss in my book together with the ordinary method used by Datasets of ADO.NET. The conclusion I came to was that the performance differences aren’t very large at all. (At least not if you make the comparison in a full roundtrip scenario. If you make the comparison within stored procedures only, the difference is a bit larger.) You should probably not use performance as a factor for which one to choose.
To test on your own, you can use the code I wrote for my presentation called “Managed concurrency control of disconnected data” at VS Live in May 2002. You find the code here.
As a bonus, you can also find code to get a feeling of the overhead of pessimistic concurrency control of disconnected data (and a lot more). As you can see, the overhead isn’t that large in full roundtrip scenarios, but when you compare it only between stored procedures, the difference is something like 1 to 7 compared to before and after values (optimistic concurrency control). But, if you have to, you have to.
:-)