After many fits and starts it looks like 2014 really will be the year network PVR becomes mainstream. Virtually all major operators are either evaluating the technology or conducting trials now that early adopters like Ziggo have demonstrated pay back through reduced CPE costs and increased ARPU.
Tivo’s leading customers including Liberty Global’s Virgin Media in the UK and ComHem in Sweden have just started pilots of the nPVR version, while in the US Comcast is now deploying its X2 cloud platform, featuring common UI and cloud PVR, with select customers in the Boston and Philadelphia areas.
Various factors are conspiring to create a tipping point for nPVR almost nine years after the first rather abortive deployment in 2005 by IPTV pioneer Fast Web of Italy, which allowed just five hours or recording per user. One factor is that cloud TV as a whole is coming of age, with early nPVR adopters also deploying some form of cloud UI. This is either rendered in the cloud as Ziggo, Deutsche Telekom and KPN are doing using Active Video’s cloud TV platform, or in the set top with the cloud used just for processing the metadata, as Comcast and other US MSOs are promoting through RDK (Reference Design Kit).
Either way, cloud UI deployment is going hand in hand with nPVR, with vendors in each category increasingly collaborating. At Dutch Telco KPN, Edgeware is supplying the recording and caching capability for cloud PVR alongside the Active Video platform for UI rendering.
Perhaps most importantly, the long standing constraint of rights restrictions is set to ease, with even the big US houses at last seeing how ludicrous it is to go on insisting on a unique copy of each piece of content per subscriber, which has prevented operators from realizing the full economic benefits of cloud storage so far.
It has meant if say 10,000 customers all record a popular program, then the operator has to record it 10,000 times, rather than just once along with pointers to that single storage location for each of those subscribers. While growing availability of catch up TV has reduced this multi-copy burden by making popular content available anyway, this is usually only available for a certain time after the scheduled broadcast and so people still use a PVR if they want to keep the program for longer.
In Europe rights are increasingly becoming available for operators to store just a single copy of content and this is allowing Ziggo, KPN, Swisscom and other early nPVR adopters to reduce storage costs and achieve greater ROI (Return On Investment) on their cloud deployments. The big US rights holders are under mounting pressure to allow single copy storage as well, not just from their pay TV customers, which they could resist, but from the seismic changes in business models and regulation.
Historically the big content owners resisted even home storage on DVRs because it disrupted their long held business models around DVB/Blu-ray distribution and advertising. But now the game is up for physical media anyway, while nPVR enables operators to block the ad skipping that has been hitting revenues as a result of home DVR storage. Network PVR brings the prospect of new revenue sharing models with pay TV operators for content owners, including impulse purchase of recording capacity, targeted advertising during playback and pay per usage.
There are also pickings to be had by enabling DVR service for more price sensitive customers that previously could not afford home DVRs. This is most relevant for MSOs given that they tend to carry the lower net worth customers. This is reflected in penetration of home DVRs being on average only half as great among MSO customers as their IPTV and DTH counterparts.
As it happens the US rights situation is currently on a knife edge with a key ruling likely to impact nPVR there soon to made over the celebrated Aereo case, which revolves around the concept of network storage, even though the service is not cloud based at all. Aereo was launched in February 2012 after raising more than $20 million from media mogul Barry Diller’s Internet conglomerate IAC, using an array of antennas the size of small coins to pick up free, over-the-air TV signals, which are then sent to customers via the Internet for $8 to $12 a month on a one to one basis. Legally Aereo’s users each lease one of the antennas, which are housed in nearby ‘antenna farms’.
Almost immediately after launch Aereo was sued by major networks including NBC, FOX, ABC and CBS. The broadcasters argued Aereo’s service was blatant theft, because it did not pay retransmission fees. But Aereo responded by arguing that it was merely providing free to air service on the same basis that customers get it from aerials on their own rooftops. The Internet is just a long antenna connection by that argument, rather than a broadcast medium. This argument hinges on a 2008 verdict by the second US Court of Appeals for the Second Circuit on Cablevision’s nPVR service.
The irony was that the court upheld Cablevision’s case and established the legal basis in the US for cloud based video storage and playback, but only if operators held a separate copy for each user. In a sense then broadcasters and content owners have been hung by their own petard by the Aereo case, because now the argument is over whether its distribution over the Internet via antenna dedicated on a per user basis, is legally the equivalent of multiple copy nPVR.
So far federal courts in New York where Aereo started operation and Boston, which followed, have agreed with Aereo that it is transmitting ‘private performances’ to individual users over their own leased antennas, rather than copyright-protected ‘public performances.’ However the networks have appealed to the Supreme Court, which agreed last week that it would consider the case. Even this ruling may not be the last word, for the broadcasters exert considerable political power and may exert pressure for rule changes if they lose the case. However if they carry out their threat to make programming available only on a paid basis, then they may end up losing revenue anyway.
To some extent nPVR will win either way. If the hearing goes Aereo’s way, then the multi-copy rule is likely to disintegrate very quickly, with pay TV operators and broadcasters combining to offer content as cheaply as possible under pressure to minimize storage and distribution costs, favoring nPVR. If it goes the broadcasters’ way, Aereo will shut down and become a footnote in TV history, but then the multi-copy rule will still be ditched in order to avoid it being invoked again. So in effect the Aereo case, whichever way it goes, has exposed the folly and stupidity of the multi-copy rule.
Meanwhile the economics of nPVR have been evolving in its favor. In the early days of home DVRs disk drive capacity was more limited and people tended to fill most of them up quite quickly and then create space for new content by allowing old stuff to be deleted. At the same time server based storage was not as cost effective as it is now, so the early nPVR deployments achieved at best only marginal cost savings given the multi-copy rule was almost universally in force with no net saving in storage capacity.
But now disk drives are much bigger so that on average a home DVR is less than half full. With cloud PVR on the other hand 10% extra capacity as headroom is quite sufficient so that storage is much better utilized, so that factor combined with the growing incidence of single copy storage, at least in Europe, means that now significant savings are being achieved.
These cost savings are being amplified by improvements in the infrastructure, achieving savings in bandwidth as well as in storage capacity. In fact storage is being traded for bandwidth through caching, which applies just as much to nPVR as catch up and live streaming.
The distinction between catch up and nPVR is becoming blurred in any case, given that KPN for example records all content and distributes to edge caches around its network, retaining all of it for either 24 or 48 hours. For content accessed during that window there is hardly any difference between catch up and nPVR. Catch up requests are served straight from the nearest edge cache as usual, while nPVR recordings merely hold references on a per user basis. Then when a given user presses play, the reference is looked up and the content starts streaming from the edge cache exactly the same as for catch up.
However customers can store content for up to a year in their nPVR stash, so even with single copy it still mounts up and requires more efficient organization of storage than catch up. Edgeware’s solution at KPN comes into play here by allowing the CDN it has built, comprising multiple distributed edge caches to operate in parallel as a single storage pool. That way less popular content can be cached perhaps just once in a single edge cache and then routed from there to whoever else wants to play it on the few occasions it is required. In such situations the extra bandwidth consumed will be more than compensated for by the storage savings. Any additional latency can also be tolerated.
For popular content on the other hand it pays to replicate it across multiple edge caches to avoid excessive unicast traffic. This arrangement mitigates one of the main problems with nPVR, which is the amount of unicast traffic generated when large numbers of users are viewing popular recorded content at around the same time. KPN, which now has 800,000 subscribers paying the extra €5 a month for nPVR, has reported this generates up to 200 Gbps of unicast traffic at peak times.
All operators face continuing growth in unicast traffic as users spend more time consuming some form of VoD in unicast mode, so it will be increasingly important to confine the traffic to the edge of the network and perhaps only the access link. It may be that CDNs will extend right to the home with caching pushed out that far, which ironically would restore the home DVR although then it would be part of the cloud.