Server-Side Stream Repackaging [Streaming Video Technologies Panorama, part 2]

There are basically two ways to sustain the extensive growth of video formats that you must, as a media distributor, serve to your different clients’ target devices : the most common answer is to choose the best in breed most-powerful encoders to prepare all the target formats during the content preparation time (see Panorama article N°1 on this topic), but you can adopt a different approach saying that you want to prepare your contents once and have the distribution part of the overall workflow take care of the repackaging and protection of the contents on the fly.

Server-side repackaging of the streams consists eventually in :
– choosing languages in audio and subtitle tracks available in the original mux (optional)
– transcoding/transrating the video content in different sizes/bitrates from a high quality video file (optional)
– applying a DRM compatible with the output format (optional)
– generating the manifest file corresponding to the target adaptive streaming technology (mandatory)
– remuxing and chunking the video data according to the output protocol requirements (mandatory)

Historically, repackaging was pushed as a quick solution for broadcasters to add iOS streams on top of existing Smooth or Flash streams. In a wider OTT/Adaptive Bitrate perspective, this alternative approach means : less files to manage in the main production workflow, less storage, less bandwidth to populate the origin servers, smaller time to contents’ online availability and easier support for new formats – shortly said, an agile path. Potentially a risky one, but quite attractive…

Hum… new approach… What are the pitfalls ?

There are some obvious limitations that come to mind when thinking of repackaging :

– devices heterogeneity : you have to ingest an input stream combination with various frame size and bitrates to satisfy all the targets’ needs and appetite for adaptive bitrate streaming. That means that you still need heavily-powered encoders, potentially hardware assisted ones. Yes, we still go back to the previous post in this series !

– intelligence : when a device requests a stream, your repackaging solution must preferably serve a stream combination which fits closely the device capabilities. So it means that you need an intelligent mechanism that will choose which versions of the original stream match the target upon registered devices profiles – a complex topic. This was Ripcode’s offer original strong point, we have to see how this has been developed under the RGB Networks umbrella. And here Seawell Networks will be the real game changer when their Spectrum server will ingest SVC streams and not only ABR as of now.

– protection : if you must protect your contents with DRMs, this can be a blocking point, as PlayReady is well covered now but not all solutions support HLS DRM (like Verimatrix) or plain AES128. And Flash is the biggest challenging target here, as no one apart from Adobe (with its own FMS server) is supporting Flash Access DRM, more than one year after the launch of  the product. Let’s hope that Adobe will urgently make the necessary steps to gain ground in the OEM market, because it also discards HTTP Dynamic Streaming implementation in most cases… Even with current DRMs support – and without Flash Access, a repackager has got to manage a whole set of DRM rules matching upon output technology, which is not an easy task.

– formats evolution : with WebM coming (and not even speaking of 3D MVC), the repackager will have to overcome the limits of H.264 as a pivot ingest format and integrate transcoding output from H.264 to WebM, preferably with massive hardware assistance as the goal is not to shift the transcoding load to the origin servers without a smart and unexpensive plan. Here Wowza seems to be the most advanced actor with WebM, but there is still a long road for its integration in CUDA. Another aspect has got to be considered, when it comes to the ingest format : mainly TS and Smooth are nowadays options. As Smooth can almost be considered as a standardized way of doing things, multibitrate TS encoders can be easily considered as highly risky choice as they leverage “proprietary” implementations, and that doesn’t seem to be a very standard approach for long-term ingest format.

For all of these reasons, you will find solutions that cover only a part of the overall features and main streaming protocols needed by the market now – but it’s still a possible path to follow, beginning right now, if you can focus on a smaller subset of features and make a smart bet on the future.

So let’s take a look at the different options available on the market now !

Doing it at home : on premise options

Anevia ViaMotion, the IPTV french (not swiss) knife
For those who don’t know the story, Anevia creators are the former creators of VideoLan(VLC) Player and students of the Ecole Centrale who builds some of the best engineers in France. For some years now, those talented guys have been developing a whole suite of server and transcoder products, targeting hospitality and IPTV markets. Now they turn their expertise in video streaming to the OTT/ABR market and provide innovative 3Screen delivery features. Their ViaMotion server does support all ABR output formats, live and VOD repackaging, and they provide one of the first WebM adaptive streaming options on the market, using server-side stream management. For now they don’t offer H.264 to WebM packaged workflow but it’s sure something will come out of their labs sooner or later.
They don’t give some much details on their website, so the best approach to gather precise info on input formats (several of them, apparently), DRM support and API capabilities is to catch them on their booth during next IBC. Don’t miss them !

RGB Networks TransAct Packager, the repackager factory
While it isn’t aimed at transcoding or transrating, relying on multi-bitrate TS input streams, the TransAct Packager is a specialized device which can be a great response to handling a massive amount of simultaneous streams. It’s clearly the best option for telcos and multichannel-delivery actors. Interesting bonus features would be language selection while repackaging and Flash Access support, which is poised to take off when HTTP Dynamic Streaming will have shown significant advantage over RTMP non adaptive streaming (RTMP dynamic streaming being a lotery option depending on CDNs’ configurations). The ability to buy the solution as a software only versus an appliance is a very good flexibility as it allows blade-type deployments. Rack power !!!

Wowza Media Server 3, running the endless innovation path
Those smart guys constantly show their high sense of market knowledge and they push the limits of what is the current edge ofstreaming technologies. Their new server version takes a high advance on all competitors by integrating challenging addons for transcoding, DVR and on-the-fly Verimatrix DRMization. Knowing the wide device coverage of this last actor, the Wowza will soon be the reference product for nScreen protected packaging and delivery – as their API choice is the widest on the market. Concerns remain on the scalability of this solution for large scale CDNs but this is at least a serious candidate for origin server configuration. They are the only one to integrate (beta) WebM transcoding and they also take advantage of hardware-acceleration for transcoding (not for WebM, as of now). We now would like them to support Flash HTTP Dynamic Streaming, but considering their harsh relationship with Adobe, it’s less than realistic to see them support Flash Access. So let’s wait for UltraViolet DRM interoperability !

Envivio Halo NMP, the pure head-end way
Envivio goes a solid way there, while complementing the 4CasterC4 with a telco grade repackaging appliance working on the basis of their own mezzanine pivot format (somehow like RGB’s approach). The two major disadvantages here are the lack of support for VOD repackaging (seems so) and Flash HTTP Dynamic Streaming. As regards repackaging of H.264 to WebM, we shall wait for Envivio response when the support for WebM becomes a reality in Quick Sync Video technology – and when it’s a real market demand… As of now, their product stands as a solid option for currently spread formats and offer interesting features for redundancy and metadata refactoring – this can be enough to fit your current needs if you can afford their price range.

Adobe Flash Media Server 4, the repackaging late-newcomer

Kevin Towes launching Flash Media Server 4 at IBC 2010

Adobe has a follower approach nowadays – being pressured to support HTTP streaming while they always pushed their RTMP protocol, and now they add HDS to HLS repackaging to their FMS features set. That’s just enough to stick to the market, as their Flash Access DRM solution doesn’t protect anything else than Flash streams for the moment. If they manage to extend Flash Access out of the box support to HLS and Smooth (not an easy task), then FMS will take back its central place in streaming architectures and could justify its price. Now it’s just another ingest server solution competing against Wowza who opened stream republishing to FMS edge servers (ouch) and the competition is gonna grow up as other low-cost initiative surface. Great expectations show to see if they can leverage the OEM market in the encoder ecosystem as they did with the Flash engine with the Open Screen Project which begins to produce results. Adobe’s current solo partition could finally hurt their universality goal if they don’t flex out their Flash Access OEM policy and find a more balanced pricing approach for FMS.

IIS Transform Manager, Expressive API power
Allowing an Expression Encoder instance to be driven as a IIS module allows for great customized workflows based on watch folders, when you know how extended and powerful the Expression Encoder 4 Pro API is by now. Add to this the CUDA transcoding acceleration and you get one of the most reasonable option to build origin servers with transcoding capabilities to fine tune output formats to different device requirements. The Transform Manager can also be piped with X264 and FFMpeg – beware ! Given the fact that PlayReady has proven a high level of versatility with iOS and Android target environments, what you get here is a highly dangerous actor. Hopefully for competitors, it doesn’t provide live repackaging features, but be warned that when it comes out, IIS will be the favourite choice for everything outside Flash outputs.

Seawell Networks Spectrum, the gifted any-to-any troublemaker
Anxiously waiting for the time when Seawell will offer the ultimate workflow SVC to all ABR techs (Andy Beach confirmed me that they are ready to release the full united SVC workflow from encoding to repackaging but they are holding it off until market demand for SVC makes sense) and definitely set the bar of streamlining production workflow, we can now already enjoy (who tested it in real life ?) their server solution already brings a nice flavor of ABR esperanto gifts, allowing for transparent input/output repackaging between major ABR technologies, for both live and VOD. Their weak point for the moment is the lack of DRM support but they are working on it right now. Nevertheless, it’s a very promising track regarding the to be confirmed SVC promises, and an interesting option for interoperability between current ABR technologies…

Codeshop Unified Streaming Platform, the triple cover-agent
Coming out from a discrete open initiative, these guys from Netherlands have managed to transform the project in a real commercial product which has already proven its efficiency in high scale streaming conditions. It’s simple, it manages many DRM flavors and it works on Win and Linux origin server platforms for a highly competitive price – a very nice partner for all standard caching solutions like Nginx. It works with Smooth streams as inputs and outputs all ABR formats including DASH for HbbTV devices. It even repackages subtitles (!) and handles multilingual streams nicely. There’s a question regarding HDS as they seem to provide only a JWPlayer/Flowplayer plugin, so it seems like it’s not really a plain standard Flash HTTP Dynamic Streaming output that they are producing – but I may be wrong if someone has an OSMF plugin working somehow. Anyway, it’s really nice to see such an approachable solution out on the market.

Online options for the cloud fans

Akamai HD Network

Server-side repackaging is the new motto of the CDNs as they want to extend their services portfolio with features that make their clients’ life easier, but in the real life they are still far from implementing it as a real extensive service. On this track we can find Akamai who deals with VOD repackaging for the moment, and Andomedia’s StreamTheWorld who proposes an end-to-end live multi-screen solution from one single RTMP input. For the latter, adaptive technologies support are unknown as of now, so this offer has been kept apart from the initial panorama. On Akamai’s side, we can imagine that this is just the first floor of the rocket which is showed right now, and that they will release full support for live repackaging, H.264 to WebM transcoding and DRMs once they are fully ready – maybe when UltraViolet goes in production through their platform. As usual with them, IBC 2011 may be the moment when previous year promises get in production – stay tuned !

Livetranscoding.com, the cloud leader
For now, the Flumotion team has got some advance on Kulabyte/Sorenson as they offer RTMP repackaging to Smooth Streaming. It’s funny to see that they still propose the good old WMV over MMS output, but I guess it still has success with the corporate clients who paid big bucks for MMS support in their proxies infrastructures. They play a good card on new formats integration with WebM support (still under the radar), following Flumotion’s early positioning on this new front. For the moment, their lack of mainstream CDNs is their major limitation, as they only offer EdgeCast CDN plans. If they smoothly get Akamai, Level3 and Limelight certifications, this would be a better situation for them, but it seems that this service goes frontally against CDN new services aspirations, so I doubt that they will manage this path. Maybe they can sell their technology to big clients wishing to build their CDNs (and there are more and more of them…) before guys like Elemental Technologies grab the market with their accelerated cloud offer.

Kulabyte Hyperstream Live, big name for a small offer
In the past, Kulabyte just outperformed all vaporware vendors with spectacular marketing announcements and few working implementations afterwards. This time the product name is still pretentious compared to its output repackaging options (RTMP and HLS), but it seems it was sufficient to convince Sorenson to work with them to complete their cloud product range with a white label live option. The interesting thing here is that you can buy a licence for managing your own server, and that you can target 3GPP too in case you need it (do you ?). Maybe there will be some demonstrations to see on Sorenson’s IBC booth, why not…

OnlineLib VCS Multiplex, the unlikely joker
With the years, those German guys have gained a good reputation in the Flash world and they are building interesting niche solutions like plugins for Rhozet covering features not yet integrated into the core program. They also develop their own PlaySmart solution of DRM for Flash, but this is not an officially acknowledged solution by the major studios. As regards repackaging, they have developed a solution which transforms RTMP streams to HDS and HLS, for live and VOD, interestingly with GPU support for the latter one. Maybe a good alternative whose evolution is to follow.

Final words

Although there is no miracle solution available yet, regarding the 5 potential actions bundled into repackaging field (listed on top of this post), it’s already possible to minimize encoders workload and simplify production workflows by deploying not-yet-very-intelligent-but-efficient solutions. Keep in mind that you certainly will be sticking one’s neck out (essuyer les plâtres), here… so good luck !

Well, that’s all for this panorama, I hope that you enjoyed reading this post and that it will be useful to your projects.

Please vote for the next subject using the survey hereafter, the next post will be released just after the IBC – hopefully with fresh intel from the ground !