Microsoft, Adobe, and DASH: The State of the Union
Time-based addressing is currently supported because it is the most flexible and resilient for live encoding and VOD, ad insertion, segment splicing, variable segment duration, multiple synchronized data centers to provide high availability and scale, etc. PlayReady DASH porting kits and DASH.js players support number and byte range addressing as well. Server support for additional addressing schemes (and On Demand Profile for byte range addressing) will depend on customer requests. We are not currently planning to provide M2TS or other multiplexed media for DASH, even though Azure Media origin servers do package M2TS on demand for HLS output, because DASH implementers have selected ISO Media as the most efficient format for streaming, most efficient for a single Common Encryption that works with multiple DRMs, and the highest quality adaptive switching and splicing in players.
Do you plan to extend the DASH features scope on Azure Media Services (such as to live packaging with DRM)? Are you already using or planning to use Common Encryption for all live/VOD streams generated?
It is currently an option to deliver DASH Media Segments with Common Encryption. Common Encryption can be applied during encoding and files stored with Common Encryption, or Common Encryption can be applied as Segments are packaged for DASH or Smooth output.
Regardless of DRM encryption, content is always encrypted with full file encryption when stored and copied in the Azure Media platform, and may be encrypted from the source with a user generated key. Storage encryption is only removed during a secure Common Encryption process.
Segment encryption is also available, which is sufficient for conditional access control, but not content protection. This is often referred to as “AES encryption”, although Common Encryption and many other encryption schemes use the US government specified AES cipher in various different algorithms. Segment encryption is the method used by HLS, and it requires decryption on arrival, so exposes clear content in players that may not be secure. Players that are given a “clear key” can decrypt the content, so some form of server authorization can be required (e.g. verify a subscription) before the keys are delivered.
Do you plan to release DASH support in the OSMF Smooth Streaming plug-in as a public feature? If yes, when and which DASH profiles will it support? Do you plan to extend DASH support in other client SDKs/Frameworks?
Smooth Streaming SDKs and porting kits enable DASH playback on XBox, Windows Phone, and most Windows and Mac platforms. The Smooth Streaming Porting Kit can be used for Smooth and now DASH streaming on TVs, set top boxes, disc players, and other devices. IE11 and other current browsers (such as Chrome) that support W3C Media Source APIs (MSE) on Windows 8, Android, etc., provide DASH playback in a web page using DASH.js or similar script. A Flash OSMF plugin is available for simple applications that run on most legacy browsers. It supports time addressing of DASH Segments, but not DRM. We can’t speculate on deployment dates for XBox and WinPhone in advance (at the time of this interview, in March 2014).
Looking at the Azure live Smooth manifests for the 2014 Olympics, we noticed that it wasn't the same format as the one used for 2012 Olympics, especially regarding the way fragments naming is constructed dynamically, with a template methodology close to the way DASH is does the same thing. Are you still evolving the Smooth Streaming format, or all efforts are now focused on DASH?
Correct: The Sochi Olympics used the currently available Azure Media origin servers with dynamic packaging capability to output multiple formats, which now includes the option for DASH format.
During 18 days, Azure Media streamed 204 live channels, automatically creating 10,000 hours of VOD content and providing high-definition streaming to five broadcasters across 22 countries in four continents from six data centers that provided 100% backup and zero stream interruptions for the entire event. This required 100TB of storage and around 500 billion storage transactions.
The Smooth format is very similar to DASH, but Silverlight is in sustained engineering support and has limited support in Internet Explorer, so DASH and browser playback will soon enable streaming to the full range of internet video devices. Smooth Streaming, HLS, HDS, etc. are necessary today to reach the installed base of players, so Azure Media Service and our player SDKs support multiple formats for a “smooth” transition.
Does the end of life (EOL) of Silverlight mean the EOL for Smooth Streaming? Until when will you support Smooth Streaming? Is there a chance that the EOL of Silverlight will be canceled, or is the switch to other client technologies like HTML5 a final decision?
Smooth Streaming will be supported as long as there is customer demand; no end of life date is being considered. Silverlight will be supported, but not improved, apart from adding DASH support to SSME and the Smooth Streaming software and porting kits. Smooth Streaming is also being used with an OSMF plugin we support for Flash players. New client technologies such as native applications (i.e. Windows 8 and iOS store apps) and HTML5 web pages are expected to eventually take over adaptive streaming playback.
Is DASH the de facto video format for new devices like Xbox One and Windows Phones?
DASH will be the preferred format moving forward. Smooth, HLS, and DASH will all be supported to various degrees. A gradual transition to DASH is likely, but that decision will be made mainly by services and content providers adopting DASH. Azure Media hosted services will make it easy for services to transition to DASH as soon as the player applications they supply or target support DASH playback.
Are there any other Microsoft initiatives that could extend DASH reach and adoption?
Microsoft OpenTech (a subsidiary of Microsoft focused on open source software development) has been active in the DASH.js open source DASH script player project on GitHub. Microsoft has been active in W3C standardization of the Media Source Extensions (MSE) and Encrypted Media Extensions (EME) that has enabled DASH playback from a web page in any conformant HTML5 browser (currently IE11 and Chrome, and implementations of WebKit). Combined with the DASH output option in Azure Media, this will make it possible for publishers of all types to reach billions of internet video devices that include HTML5 browsers, Google Chromecast dongles, connected TVs, game consoles, disc players, mobile devices, PCs, etc. with a single web page containing a proven but customizable DASH player.
Adobe: Picking Up Steam Towards DASH
In early 2012, Adobe's Kevin Towes announced that Adobe would be supporting MPEG-DASH across all Adobe’s online video product range. 2013 led us to think that DASH support would be restricted to Primetime and that the actual support for DASH would come in 2014. Both these scenarios are turning true, as confirmed by director of internet video product and experience Ashley Still product manager for Primeteime and Rick Wisher just before NAB, where Adobe announced support for DASH in Primetime.
Since the first announcements in 2012, the ecosystem has been waiting for a long time for Adobe’s implementation. What is the status of DASH integration into your product line?
First let's say that DASH is additive for Primetime. We're fully committed to meeting all of our customers with the streaming technologies they want to use, with the protocols they want to use. We are here to enable their business. As we start to add DASH we're not going to take away support for other streaming platforms.
The other core message we want to deliver about DASH is that we will make it easier for our customers to deploy. Our DASH implementation is going to be consistent across platforms. We support a lot of different platforms, we're now starting to look at the HTML5 browser space, lots of digital home devices like set top boxes, televisions, gaming consoles. Where our DASH implementation will be really superior is that we'll be able to play the same content back across devices, and this is really the same message we have with HLS.
The final point is that all the workflow features, like ad insertion, analytics, ABR, all the different flavors of close captioning & subtitling that need to be supported, or failover, or blackout workflows - all the great things that we've done to make our customers successful in 2012 and 2013, we'll also support that across all the players in DASH.
The Adobe Primetime video workflow
So that's how we are looking at DASH now: it's just another protocol, hopefully a widely adopted protocol that works across a bunch of different platforms. Our value is that we'll make it easier for our customers to transition into DASH and we'll support the same feature set that we supported in the previous versions of our player.
Does it mean that your development efforts are still ongoing with HDS? Or can we say that somehow DASH will replace HDS?
We don't have any plan to sunset any of the protocols that we support today. Of course as we move forward as it makes sense for our customers' needs and where they're making new deployments, that's where we will focus the most resources. Again, DASH is additive; it's not in replacement of anything. Our mission is to enable television, and we consider DASH, HLS, and RTMP as tools to accomplish that goal of helping our customers revolutionize the television experience and engage with consumers in new and exciting ways.
We think that DASH is a natural evolution point for customers that are on HDS today, but we have no plans to end of life anything, RTMP or HDS. But certainly we see in DASH a great promise of support across many different platforms. We think DASH is great, we hope that it will be supported everywhere, and if it's the case it truly will become that single protocol. But DASH has got to solve the problem and become the de facto standard on its own yard.
You say "DASH will be supported by Primetime." But for the sake of precision, can we also say that "DASH will be supported by Flash Player, outside of Primetime?"
Right now our plans are purely to implement DASH into Primetime players. It won't be available in the mainstream Flash Player, but it will be available everywhere the Flash Player is available. When you think of Flash Player as a whole, there are two separate video pipelines in Flash player today. One is the one that is 100% free and that anyone can use: that pipeline will not support DASH. There is another video pipeline in Flash Player that is Primetime, but what that means is that you have to be a Primetime licensee to have access to that video pipeline, and that is where we are focusing our ongoing development, overall, for video. We changed our business model, which used to be completely free platform with revenue on Flash authoring tools. We now charge directly for the use of our technology whether it's mobile players or the Primetime video pipeline, which uses Flash Player for distribution. So absolutely, we'll be able to proliferate DASH support over 99% of the internet—through Flash Player—and you have to license Primetime to access that pipeline. That is a practical business model so that we can support our development efforts with building all of these new technologies.
What’s your updated roadmap for commercial launch of DASH support into Primetime?
It’s later this year. Right now we have been looking at a schedule that takes us sometime in Q3 to really start to deliver core player features across platforms. We’ll have support for a lot of the use cases we support today. Definitely this year.
And which DASH profiles will you be supporting when you release?
So far we have been looking at the DASH-264 spec. We're part of the DASH Industry Forum, so members of our team have been working with the larger group to really figure out what that final spec looks like, and that's what we'll be supporting. Also we'll be working with our customers to figure out what their needs are as well. That's where we are really marching towards right now. In the future as it becomes widely adopted we will also support DASH-265, as we want to support to UltraHD and the highest encryption standards.
Will it be possible to apply Access DRM over Common Encrypted DASH video fragments alongside any DRM compatible with Common Encryption?
You already can do that, obviously not for DASH, but as we are part of the DECE consortium, we follow one of DECE requirements, which is to support Common Encryption, so we support that already today. There is no reason why we would break that support for DASH. So yes you will be able to use the Common Encryption.
And do you have plans to support DASH on iOS?
We don't have current plans to support DASH on iOS because we leverage the AV Foundation and Apple has not announced any plans to support DASH. But as soon as they do, we'll be there as well.
Do you see major traction for DASH in the market?
There's definitely been a lot of demand for DASH. DASH is gaining a lot of momentum. It's really where a lot of the industry is moving. It has our full attention and the attention of a lot of our customers too.
The promising format is already splintering. Cisco, Akamai, the BBC, and Qualcomm offer hope for a standardized, unified approach.
At NAB earlier this month, the DASH Talks delved deep into the ways that the standard calls for conformance and allows services like HbbTV to move forward
MPEG-DASH is slowly but surely becoming the main competitor to HLS, driven by adoption by major players and intrinsic strengths. Here's who's using it now, who's going to be soon, and what challenges still need to be addressed.