In the last days my team has been working in its first Facebook application. For convenience, we’ve chosen to develop it using iFrames, but we took this decision without thinking too much about other possible benefits or inconvenients. When approaching the final development phases, we’ve faced some performance bottlenecks that made me doubt about our choice. But after making some research, I’ve found this interesting article that convinced me we took the right path:
At the end, our real problem was not the iFrame choice itself, rather some bad designed pages that made a lot of unecessary calls to the Facebook API. After correcting this, the performance problems were eased.
UPDATE (April, 2011): Some months after this post was written, I see we really took the good decision, since Facebook has begun to deprecate FBML.