How to Implement Drill-down Functionality in Telerik’s PivotGrid for ASP.NET AJAX Part 2

This is the second post of mine, which elaborates on how to implement drill down functionality in Telerik’s ASP.NET PivotGrid control. In the previous post I demonstrated how to display our Popup Dialog control containing a Telerik Grid with detailed information about the PivotGrid’s data cells. Today I will show you how to optimize the performance of the example provided in my previous post.
The drill down demo from the first part is great, but it is not perfect. It has some weak parts and we can improve them. Here’s how.

Kill the LinkButtons

First, we render LinkButtons and if the PivotGrid has thousands of data cells, thousands of LinkButtons will be added. However, with so many controls on the page the ViewState will be gigantic. Also, the rendered HTML will be very large and the browsers can be slower when rendering it. So let’s kill the LinkButtons.
But wait a second, if there are no LinkButtons, where will the user click in order for the window to pop up? The answer is simple - on the data cells. We can attach a client-side onclick event to every cell and fire an AJAX request or command to the server.  Also, instead of setting the select query arguments as LinkButton CommandArguments, we can pass them as a parameter of the onclick event handler:
Then on the client side we can fire an AJAX request with the related to the data cell select arguments:
If you inspect the ViewState of the page after this change, you will see that it is twice smaller than before the changes. I think this is our first victory over this example.

Eliminate the Millions of Checks for Each Data Cell

In the previous post I talked that we need to adjust the row and column indexes by calling AdjustRowAndColumnIndexes:
However, in the best case each data cell will pass through three checks, in the worst case - more than ten. This is a lot of “if” conditions. If there are 10,000 cells, we will have more than 50,000 checks in order to differentiate each cell. Can we decrease the number of checks? Yes we can, but how? If the cells belong to a matrix, we can cache the first row column indexes and the row indexes of each row’s first cell. The good news is that data cells belong to a matrix. So we can cache the indexes of the fake cells. First let’s cache the column data cell indexes:
Then we can do the same for the first cell from each row and apply the result to each cell from that row:
With the described changes we decreased the number of data cells checks considerably. But still there is a weak place left - the arguments passed to the client. 
I described in the previous post that for each cell we pass an argument with the following signature: “columnName~value;columnName~value”. For example: 
TransportType~Microvan;Company~Advanced Services Ace;Country~Bulgaria;City~Sofia
however this is a big string and for almost  every data cell there is a repeating part:
TransportType~Microvan;Company~Advanced Services Ace;Country~Bulgaria;City~Sofia
TransportType~Microvan;Company~Continential World Corp;Country~Bulgaria;City~Sofia

Shorten the Strings

Unfortunately transferring very long strings to the client is not a good idea, so we need to perform some optimization, but how? One of the answers here is to replace each string with a number. In this case we can create a key-value pairs of a number and a real string and these numbers will be sent to the client. We can keep key-value pairs in a dictionary stored into the Session:
Then into the GetCommandArguments method we can build the number arguments which will be sent to the client:
Now instead of sending 
TransportType~Microvan;Company~Advanced Services Ace;Country~Bulgaria;City~Sofia 
to the client we just sent something like this:
'0~1;2~3;4~5;6~7'

Conclusion

You can see the example of the demonstrated drill down functionality into our SDK. You can download it and think about other optimization techniques which can improve the performance of the Drill Down functionality. I will be glad to hear what ideas for optimization you come up with.