Date: 5th July 2010
Speaking as web journey monitoring people, there are a number of reasons why a move away from Flash for delivery of streaming video, to a slightly more standards standard like HTML5 is attractive.
In a nutshell, writing user journey scripts to benchmark or load test the performance of a video streaming site (or any non-trivial web site for that matter) can be quite tricky when you have to either emulate ‘doing what the browser does’ – or else control and manage the browser to do it – and the nature of Flash’s streaming with RTMP under the bonnet, and the layered intricacies of how folks add authentication, encryption, DRM and etc: can make for a major effort.
Whereas the promise of a standard that is open and a standard from the start, could make that much easier building tests.
But it’s interesting that Youtube engineers have come out this week and said they’ll not be dropping Flash for a while – or ever for certain cases: despite them being early adopters and advocates of HTML5 – they already have their own YouTube streamer for HTML5 for example.
They say only Flash can handle all the range of things they need like: Encapsulation and Embedding, and they like the fine control over buffering and dynamic quality control…. some of the features that makes Flash video harder for us poor performance guys!
But maybe organisations with less demanding streaming needs than Youtube, will find it easier to jump ship 100% more readily.
Of course there’s been a bit of a debate online as to the relative merits of Flash and HTML5 for streaming – HTML5 is new and relatively unproven and so on. If streaming has a big role on your site, confidence in the protocols you are using is crucial: and so a move to HTML5 will not occur overnight, even if it did tick all the boxes of web performance.
And claims that HTML5 would use a lot less CPU for users, has not proven wholly conclusive (depends who you ask…).
And of course so far, Microsoft say IE won’t support the video tag from HTML5.
Maybe ending this post with IE is an appropriate down note: despite any desire for sensible engineering approaches, the complexity in web monitoring and load testing of video streaming seems unlikely to go away anytime soon.