Thanks Cathi,

I wonder what differences, if any, there'd be using Framework version 1.1. Does he plan on testing that do you know?

~~Bonnie

>Bonnie,
>
>The test used a machine with Windows 2000 Server, SQL Server 2000 on the same machine, and version 1.0 of the Framework. He ran the test using the Northwind database and retrieved results using 10 records, 50 records, 100 records, and 500 records. The percentages below are an average of the test results for each number of records retrieved. The first ten tests run were thrown out to remove any skewing of results for the first hits of the application. The test results were retrieved by creating a subclassed page class that had the Init and Unload methods overriden. These methods tracked the page execution time. He used API calls instead of the .NET timer since he needed better accuracy. To run the tests, he used Application Center Test. I'm not sure what he used for the number of hits but it was the same for all the tests to make the environment consistent.
>
>>Hi Cathi,
>>
>>Some interesting numbers. Do you know any details about how he ran his speed tests? You know, things like how many records he was accessing, or if the server was under a heavy load or not, things like that. Just curious ...
>>
>>~~Bonnie
>>
>>>Hi All,
>>>
>>>I was talking with Stephen Walther tonight (author of ASP.NET Unleashed) and he has done some testing of data access with ASP.NET using different scenarios. I thought I would pass on his results.
>>>
>>>Using a DataReader instead of a DataSet is 16% to 50% faster, depending on the configuration of the DataSet like enforcing constraints.
>>>
>>>Using an ArrayList resulted in the same speed performance as a DataReader. Using an ArrayList gives you the added ability to have static representation of the data whereas the DataReader does not.
>>>
>>>Using the SqlClient set of classes is 52% faster than using the OleDb classes when accessing SQL Server.
>>>
>>>Using C# code to access data via a DataReader object is 4% faster than the VB.NET equivalent with VB.NET Option Strict set ON.
>>>
>>>Using a SQL Server Stored Procedure was 3% faster that using a dynamic Select statement. The reason this difference isn't greater is SQL Server's ability to compile dynamic statements and store them temporarily.
>>>
>>>Accessing columns in a DataReader by ordinal is 14% faster than by name. Using the GetString method is even slower by 11% than accessing by name.
>>>
>>>There is a 4% penalty of accessing DataReader column names using the wrong case. ADO.NET has to go through an extra step to resolve the inproper case.
>>>
>>>Using an Inline DataReader is 63% faster than using a DataGrid. A DataGrid has a lot of overhead associated with binding the data.
>>>
>>>Using a DataGrid with a DataReader is 6% faster than using a DataGrid with a DataSet.
>>>
>>>Using DataGrid Templates is 24% slower than not using templates.
>>>
>>>If you use DataGrid Templates, databinding using explicit casting is 9% faster than databinding using the Eval function.
>>>
>>>Explicit casting:
>>>

>>>< %# ((DbDataRecord)Container.DataItem)["ProductName"]
>>>

>>>
>>>Eval Function:
>>>

>>>< %# DataBinder.Eval(Container, "DataItem.ProductName")
>>>

>>>
>>>Using the DataGrid's ItemDataBound event handler to databind controls is 46% slower than using explicit casting shown above.
>>>
>>>Caching a DataSet to use in a DataGrid is 72% faster then not using caching.
>>>
>>>Using a cached DataSet is significantly faster (he didn't have the exact numbers calculated) than using a DataReader.
>>>
>>>Using AutoGeneratedColumns for a DataGrid is 24% faster than specifically selecting columns.