Recently I posted about Timestamps and CommandBuilders and I got a well informed reply by Luciano Evaristo Guerche concerning a related approach of using BINARY_CHECKSUM in SQL Server to do the same thing. I think Luciano's response means to say that if you can't use Timestamp's in the database (like you don't have control over the schema) then BINARY_CHECKSUM is an improvement over the brute force concurrency that CommandBuilders do by default.
I thought Luciano was right, but I wanted to prove it out. I ran some tests using a Typed DataSet and the Northwind Customer table:
In general the results were what I thought. The brute force performed worst, the BINARY_CHECKSUM was a performance boost, but not to a great extent; and the timestamp showed considerable improvement. Here is a chart of my results. The number of rows updated corresponds ot the batches (e.g. 5 iterations x 91 rows = 455, etc.) and I have calculated the row/second response time from each result:
Let me know if you have any questions...
|Building a Web App with ASP.NET Core, MVC6, EF Core and AngularJS (New)|
|Building a Web App with ASP.NET5, MVC6, EF7, and AngularJS (Retired)|
|Best Practices in ASP.NET: Entities, Validation, and View Models|
|Front-End Web Development Quick Start|
|Lessons from Real World .NET Code Reviews|
|Node.js for .NET Developers|
|Implementing ASP.NET Web API|
|Application Name||WilderBlog||Environment Name||Production|
|Application Ver||22.214.171.124||Runtime Framework||.NETCoreApp,Version=v1.0|
|App Path||D:\home\site\wwwroot||Runtime Version||.NET Core 126.96.36.199|
|Operating System||Microsoft Windows 6.2.9200||Runtime Arch||X86|