Silverlight: What you Need to Know
- Categorized in: Distributing with Silverlight
But let's start with some background. Before 2004 or so, Microsoft ruled the streaming video world. RealNetworks, the early leader, had the audacity to charge for what Microsoft and Apple were giving away, and was retreating from a battle that it couldn’t win. Apple never really challenged in the non-movie trailer space, mostly because their primary codec, Sorenson Video 3, couldn’t compete with either RealVideo or Windows Media. And Flash? Well, that was a joke, a great design tool, but the video was out of synch with the audio and looked awful.
Then, in 2002, Macromedia added support for the Sorenson Spark codec, with VP6 support coming in 2005. That gave Flash equivalent or better quality than Windows Media, along with vastly superior configurability and support for the Mac and Linux, where Microsoft lagged.
By 2007, Flash had displaced Windows Media in most high profile broadcast web sites, and was starting to make inroads into the Rich Internet Application market. To stem the tide, Microsoft released Silverlight, which added player configurability, a Microsoft developed Macintosh client, Linux support from Novell and a rich back end accessible via .Net, C# and other compatible languages.
You can read about how Flash and Silverlight compared in Distribute Expertise: Flash vs. Silverlight. The net/net was that Silverlight was great if you had lots of legacy WMV content to stream, or plentiful .NET and C# programing resources, but that the sub-25% (at that time) installed base limited its use for most web sites that couldn't count on a potential viewer downloading a new plug-in to view their content.
Silverlight was caught in the classic chicken and egg problem. That is, until player penetration achieved critical mass, few web sites could use Silverlight, but until web sites started using Silverlight, it would never acquire that critical mass.
To break the log-jam, Microsoft focused on Silverlight’s HTTP-based Smooth Streaming technology, which is primarily used for large scale live events like Wimbledon and the Olympics. Specifically, Microsoft framed the Flash vs. Silverlight discussion as Smooth Streaming vs. Adobe’s RTMP-based Dynamic Streaming. Of course, if you’re not producing high profile live events, this positioning is irrelevant, as I commented in my NAB Millimeter blog, Microsoft and the Art of War.
I took a deeper look at the HTTP vs. RTMP streaming issue in Streaming Gets Smarter: Evaluating the Adaptive Streaming Technologies. As the article details, while HTTP is (perhaps) theoretically more efficient in large event streaming, none of the CDNs that I spoke with charged more for Flash streaming, which muted Microsoft’s argument. The article also discusses Apple’s HTTP-based streaming for the iPhone and QuickTime X Player.
My frustration with Silverlight’s lack of penetration – and resulting irrelevance to most producers – reached a peak when I taught a summer course at Stanford, and spent 30 minutes describing what Silverlight is, and then ten minutes explaining why very few web sites could use it. This led to my commentary, Reflections on H.264 and Silverlight from a week at Stanford.
Then, Microsoft's Smooth Streaming-based strategy started to pay off, with a high profile announcement from NBC about using Smooth Streaming for this year's NFL season and the upcoming Winter Olympics. I wrote about the significance of this announcement in Smooth Streaming - Silverlight's Trojan Horse. While still unsure about whether HTTP was more efficient to stream than Adobe's RTMP, it appeared that Microsoft was effective at creating that impression, and as we all know, perception is reality. This led to my prediction that Adobe would announce an HTTP-based streaming technology within the next twelve months - if only to blunt Microsoft's messaging.
Nine days later, Contentinople reported that Adobe would include an HTTP-based streaming technology with the Flash Media Server 4 (Adobe's FMS 4.0 Will Support HTTP, Sources Say), though Adobe would neither confirm nor deny this report. Yes, that faint sound you hear is me patting myself on the back.
I interviewed NBC-SVP Perkins Miller on the Silverlight announcement for StreamingMedia magazine. You can read the interview here, as well as download a podcast of our discussion. Interestingly, while Miller obviously felt Silverlight was entirely appropriate for his exclusive, high demand content, he did comment “I think it would be perhaps a different story if you were trying this on a day to day basis in the news business.”Translation? Unless you're broadcasting Sunday Night Football, Silverlight's low penetration rate is still a concern.
My colleague, Dan Rayburn, did some follow up on the HTTP vs. RTMP efficiency issue in his post Level 3 Comments Of The Future Of HTTP Based Streaming. While Level 3 did comment that HTTP was cheaper than other protocols, they also reported that this wouldn't be reflected in their pricing in the short term.
Where does that leave us? For most web sites, nothing's changed since I left my class at Stanford, since Silverlight's penetration is still far too low for sites without "exclusive content." Still, Adobe is clearly playing catch-up in the live event streaming technology space, and it will be interesting to see what they announce and how they announce it. A strong HTTP-based product offering would affirm Microsoft's technical approach, but could also provide an easy alternative for Flash developers who really don't want to switch to Silverlight.
That said, Adobe is far behind Microsoft and others (like Move Technologies) in deploying their adaptive bitrate streaming technology, which has become a high profile battleground. They've also been strangely quiet on the PR front. Given the importance of Flash to Adobe's overall strategy, I expect a significant rebound in both areas, but only time will tell.