I am currently crafting together a pure html/jquery app to run on the Amazon S3/Cloudfront platform.
I am also using Kendo UI beta and will roll my app out around the time v1 is a month or two old.
I just discovered a giant problem for companies like myself planning highly scalable/ big traffic sites that use S3 for file storage.
Allow me to explain as this could be a HUGE HUGE selling point for the Kendo UI framework for those apps written against S3 storage.
My future users will store files on S3. I am using Kendo Grid to display that file information.
Now here is the issue.
Amazon REST API lists a MINIMUM of 1000 files per REST API GET.
I did a quick fiddle and that is exactly 300kb in bandwidth.
The data cannot be cached or new adds/edits/deletes won't update the grid.
So when using a jquery ajax get call to refresh the grid one must add an arbitrary query url parameter based upon date or the grid will not change in IE.
Ok... so now imagine that everytime you just want to read one file's worth of information from S3 you are forced to accept 1000 at 300kb.
Now multiply this by a userbase of anywhere from tens of thousands to millions of users creating data files all day and night.
In my business plan our architecture would have run in to tens of thousands of dollars in bandwidth because Amazon S3 will not allow you to get the record count without listing every file, nor will they allow you to do the equivalent of
select top 1* from S3 order by lastmodified desc
nope... gotta take 1000 keys for 300kb each CRUD operation.
So this is putting millions upon millions of dollars in their pocket based upon user bandwidth.
And then to create a sortable grid means you have to keep reading all the file info. So if you have 5000 files to sort them that means you have to load 5000 files every crud operation or your sorts & filters are incomplete.
Now here is where Kendo can save the IT community in the S3 cloud MILLIONS!
Create a SIMPLE methodology for pulling the amazon S3 file info ONCE per user session and storing via HTML5 local file storage.
If you can figure out how to use say a cookie to keep the two in sync, in other words, load once then update both S3 and local file storage to stay in sync you could have a serious compelling reason for every single app developer needing to present S3 file information to purchase your very promising UI.
I saw your post on http://www.kendoui.com/forums/framework/data-source/html5-localstorage.aspx and I am going to "roll" my own because my app needs to handle millions of users and local file storage coding promises to drive my overhead down to near zero by reading S3 files per user per session only ONCE rather than once per crud operation.
Just an idea! :))