Rethink TV’s “Multicast-ABR Market Forecast 2022-2027” is available HERE
Faultline has come to understand that Comcast never actually deployed multicast-ABR (M-ABR) technology, in a project originating in 2015 from its VIPER (Video IP Engineering & Research) wing on its Arris-supplied CPE.
The prevailing wisdom has long been that Comcast had developed this M-ABR system internally, and – as Comcast has been infamously tightlipped – we assumed the deployment was plugging away in the background. Of course, we have knocked Comcast’s doors numerous times to no reply, but – unfortunately – it says a lot that Faultline knows that the best way to get an answer out of a company these days is to write the upsetting article first, then wait to get someone on record for the follow-up piece.
Speaking to a number of sources and supported by the complete lack of public confirmation of the project’s deployment or current status, we have joined more than a few dots. If you start at a fresh Google page and go looking for evidence that Comcast’s M-ABR project ever existed, a lot of stones need to be turned over, and you will find no affirmative answer.
The most concrete proof we have of Comcast’s M-ABR development for its network, in the public domain, is a Streaming Media West presentation from 2015, where Comcast VIPER and Arris both sat on a panel titled ‘Benefits of Deploying Multicast-Assisted ABR Within an Operator Network.’
Here, Comcast describes how the system works, where the client gateways (the cable modems) run a multicast client. The client inspects HTTP traffic, looking for linear streams, reporting back to the multicast controller and server in the network. The clients are essentially voting on what streams should be encapsulated by the multicast server, which essentially intercepts unicast streams between the CDN and the client, and bundles them up accordingly. The system uses NACK-Oriented Reliable Multicast (NORM), developed inside the US Naval Research Labs.
The tone of the presentation is that this is a project that has been developed and is due to be implemented. At the time, it sounds like it is live, as it is discussed in the present tense, and so it sounds like live trials of some kind were completed. When discussing the growth of HTTPS streaming, which encrypts packets in transit, we hear the phrase “we’re going to have to figure it out on the cable modem if we’re going to continue to use this strategy,” which seems about as categorical as it could be.
Arris discusses the technology in the same manner, with a focus on audience viewing patterns as a function to leverage, by “effectively pushing the CDN edge into the homes.” In a 1,000-home sample, at peak viewing time, 1,400 streams need to be delivered. However, there are usually about 300 channels’ worth of content being requested, suggesting significant bandwidth savings from M-ABR.
However, from this point on, there is scant mention of M-ABR from either party. Searching “Comcast multicast ABR” leads you straight to the 2018 forecast from our research arm Rethink TV, as the first result in an incognito session, with the first page containing a couple of press coverage entries for the forecast too. An NCTA Technical paper from a Comcast staffer, who is currently a Senior Fellow (SVP), is there too, co-authored by a current Cisco staffer, who worked at Arris in the right time frame for the initial project development, before moving to Imagine Communications in 2014. Again, Comcast and Arris are found in the authors list of a number of CableLabs papers.
Searching LinkedIn for the same phrase turns up a wealth of R&D types, clearly demonstrating that the work was done internally, as it is stated on their profiles. In 2017, the CEO of the National Cable Television Cooperative said that the expectation was that all new Comcast video customers would receive a fully IP-based service, which aligns with the M-ABR expectations.
Once you add the term “from Arris” to your Google search, you will again find that same Rethink TV forecast at the top, but shortly after that, you will be led to the Streaming Media Blog, which is how we in turn discovered the panel discussion. You will also find mention of the Arris MG2400 DVR, in 2014, which supported multicast streaming, as well as the then-in-development DCX3620 headed gateway.
The latter was a successor to the design Comcast used for its XG1 gateways, which rather suggests that M-ABR was on the cards for the design. Searching for that model number now draws a near complete blank too, and the CommScope website draws exactly one result for M-ABR in its current products, which we think is likely because the page lists both Multicast IPTV and ABR streaming video as subsequent bullet points. It looks like M-ABR has been scrubbed from the CommScope site entirely.
Who knows how many blog posts, press releases, and updates have been removed from the searchable internet since 2015. We can probably forgive Arris, which has dealt with two mergers since that time, for losing track of its archive. Comcast, however, has no such excuse.