I would be very interested to see a simple paging vs paging test! I think paging is the obvious solution for any dataset large enough that the size creates a performance issue, whether the paging is coded from the front- or back- end. I imagine enough developers would be interested that paging code would be easy to find.
On the other hand, if Flex is crazy fast with data, I’d like to see how it handles situations where you don’twant paging, and you want5000 rows. (i.e. graphing large datasets)

@Michal – Paging is not something specific to Flex. It’s a technique that is widely used for large data sets. Flex (as well as Open Rico) support transparent paging in datagrids. Other frameworks utilize a user driven paging paradigm. Either way you are right that paged data is a different paradigm than non-paged data and I should make this more clear in the results. Thanks for the feedback. This will be at the top of my list for the next version of this benchmark app.

Thank you for updating the tests. I still think that paged Flex bandwidth test (which _does_ transfer less data) result should be separated from other results and marked for obviousness, or removed from the test whatsoever — it’s comparing apples and oranges. Think about it, wouldn’t you raise an objection if I posted results showing that Ajax/Json used 50 times less bandwidth than Flex?

There’s a chart that makes it look like flex uses a tiny fraction of the bandwidth used by other techniques. But when you read more of the details you find that the flex test just doesn’t load all the data.

What–other than an attempt to deceive–would cause someone to make a chart comparing flex’s performance on partial data against other technologies’ performance on a full data set?

I really think you should re-think how you’re testing.

Well, at this point, I wasn’t so wrong 8)

If I understand correctly, the author knowingly lowered the transfer time and bandwidth results for paged Flex 50 (fifty) fold. He writes “I ran a test using 5000 rows” whereas the results for paged Flex are given for a transfer of a 100 rows.

In comments above, he writes “JSON is usually only slightly smaller than XML because all you really save on is the close tag.”, which in itself is only a half-truth at best. What I’d like to know, though, is why does the “slightly smaller” JSON transfer take twice the bandwidth of XML?

I’m sorry, but this benchmark seems to be paid for by Adobe.