The MetaMessage in Mobile Media

While sitting in editorial meetings at 4G World last week, it became clear to me that a new wave is coming. 

The new wave has something to do with the interconnection between mobile and media; where the content delivery networks and service providers transfer. It is taking us beyond the concepts of offload and toward a new era of connectivity. 

For my friends who are all about the Internet, this is not about the best effort but rather the aggregation network. In a mobile world the last mile is harder to predict.

The communication between users devices, the carriers evolved packet core and the handoff to the backbone has several places where traffic can be managed. Listening to the briefings I feel like the new wave is like the elephant and the companies are the monks each touching a part.

Seamus Hourihan of Acme Packet sees most of the elephant when he talks about Session Delivery Networks as compared to Content Delivery Networks. He starts with the perspective that end-to-end (E2E) communication has a real-time component to it. Managing the sessions, as opposed to distributing the content, has better quality of service, trust and regulatory compliance.

From the viewpoint of telecom, he makes a lot of sense. However, the Internet is now taking over Broadcast Television faster than it took over the PSTN, and I am not sure that end-to-end is the same guiding principle that it used to be. (We will also ignore the irony of B2B SIP user agents, talking about E2E.)

If you look at the Internet’s backbones, it is no longer the case that one size fits all. The CDNs now dominate the traffic, but they are not designed to handle the Real Time Communication (RTC). 

In some ways this is a problem of Telecom meeting Data. Telecom people see the world as being about sending and providing a full experience. The truth is, particularly with voice, they provide a sample of the voice and not the whole experience. It is clear that we have missed the point of best effort. As Jim Gettys has pointed out the requirements for TCP are very different from those of UDP.

Saying it for the layman, “Watching” with TCP, is different than “Talking” with UDP.

Or putting it in Cool Hand’s Luke’s terms, “what we have here is a failure to communicate (correctly).” And when we deal with mobile, the issue becomes one of matching the media with the transport. Mobile is by nature a session-oriented experience, so a blend is needed. The wireless operators would tell me they do all of this, however, we are rapidly getting to the point where over the top and interconnection of VoIP with carrier voice (regardless of technology) is becoming more common. Also consider the fact that carriers have yet to get themselves integrated into the video calling opportunities. 

I want to call this merged media mayhem “media handoff management.” 

Seamus is right that our future is in sessions, but the principles of end-to-end are going to be radically changed. 

• If we get ‘Watching’ right, the TCP buffers will be optimized for some loss and for cached distribution (and please G-d some form of multicast).
• If we get ‘Talking’ right, the UDP will go end-to-end only when both parties are in real-time, otherwise voicemail should be an FTP originated from the calling party and not a waste of the network.
• If we get ‘Mobile’ right the on load/off load handoff will bond caching with end-to-end sessions as appropriate for what is being communicated/displayed.

Since mobile is our future and sessions are essential, Seamus and Acme Packet are seeing the problem the best. However, based on the traffic on IP networks we are still very blind monks.