I have a question. I have a rather large dataset (~ 1M records) containing track and trace log messages. The average number of messages per T&T code lies around 20 records. I have a page on which these log messages for a specified T&T code are displayed using a Kendo MVC Grid, without pagination (and no pagination is intended here). As the table is getting bigger, the web app response becomes slower and slower.

I have done some profiling, and it turns out that the Json(result.ToDataSourceResult(request)) invocation in the Get ActionResult produces two queries. The first one is a (very slow, ~ 18 s) query which seems to take care of the 'Total'-property in the DataSource result. The second one is a much faster (~ 150 ms) query getting the actual records.

Since we don't need pagination on this specific grid, I don't really understand why the total count is needed anyway. Can't Kendo be smart enough to skip getting the total count when we don't need it?

I have searched the internet, and it seems like there are more people experiencing the same thing: http://stackoverflow.com/questions/15843703/kendo-mvc-todatasourceresult-extremly-slow-with-large-iqueryable
The reactions there suggest to create a custom binding, but I would really prefer not to enter that path..

Is there a solution for this?

Hi,

Currently the Count() is executed always. You basically have two options:

• Modify the implementation of the CreateDataSourceResult method (src\Kendo.Mvc\Extensions\QueryableExtensions.cs) and build a custom version of Kendo.Mvc.dll
• Implement custom binding

On a side note you mention that you have 1M records but don't want to implement paging. How do you plan to avoid selecting all items in memory?

Regards,
Atanas Korchev
Telerik

I have 1M records in the database table, but the records are filtered to an average of ~20 records.
I will take a look at the custom binding possibility. However, I would very much like an option to skip the count.

I have partly fixed the problem by pre-filtering the specific records.

The two solutions you suggested are both not really acceptable for me. I started to code the custom binding, but realized I had to do the complete filtering myself. Which is not a nice thing to do, for a ~20 field entity. The pre-filtering option did the trick, at least for this part of the application.

I still think it would be very nice to have the option to disable the count, or to do be able to implement a custom count action For some data sets, you just know that there are a lot of results, and you don't really have to know how many. It suffices to just press 'next page' until you have found what you're looking for. Kind of similar to a search engine like Google when you search for 'internet' for example (although they do give you a raw estimate of the results).