Virtual scrolling with different pageSize to DataSource

[Rowan]

Hello,

So Grid virtual scrolling is a great concept and allows us to show large amounts of data with good performance. The problem we have is that it couples two unrelated performance bottlenecks into a single pageSize variable on DataSource:

1. Backend/Remote fetch performance and processing.

2. Client side HTML rendering performance.

Generally speaking we can fetch a large number of items from the Backend without hitting any performance bottlenecks. This makes for the most efficient use of bandwidth (smaller HTML request overhead) and a better user experience (don't keep hitting the load bar). The problem is that if we tune a larger value for pageSize to make (1) work better, then suddenly we start hitting Client side rendering bottlenecks in (2).

In an ideal world the pageSize used by the Grid for display virtualization would be an independent value that could be tuned differently to the pageSize of the DataSource. Is there a way to achieve this in the current library?  I was thinking perhaps I could somehow wrap a remote DataSource with pageSize A in a local DataSource with pageSize B, and then set the local DataSource on the grid.

 

thanks,

Rowan

[Nikolay Rusev]

Hello Rowan,

 

I'm afraid that this is not supported. With current implementation pageSize is directly related to rendered items in the Grid and how DataSource calculates how many items to request.

 

Regards,
Nikolay Rusev
Telerik

 

 

[Rowan]

Hi Nikolay,

One thought I had was that there is already an undocumented prefetch method on datasources that populates the _ranges cache. It seems that we could use prefetch to get a large batch of data from the remote transport (without rendering it), and then have a small pageSize that would just get it's data out of the _ranges cache when refreshing the view (rather than hitting the remote transport again)?

 Does that seem like a reasonable approach?

 thanks,

Rowan

[Nikolay Rusev]

Hello Rowan,

 

We are not encouraging using internal, undocumented API as it might lead to unexpected behaviors. The _rages cache is used to for getting pages with are already fetched as virtual scrolling pre-fetches pages ahead of the time for more smooth scrolling experience.

 

Regards,
Nikolay Rusev
Telerik

 

 

[Umair]

This is something I could really use. Any updates on this? Is pageSize is coupled with both Client Side rendering & server request?