Flash memory caching, peer-to-peer protocols and video search all have role in on-demand future

Philip Hunter investigates the challenges presented by increasing volumes of on-demand content and asks what role P2P could play in making ‘long-tail’ content available across IPTV networks

The greatest challenge for IPTV has little to do with the IP protocol itself, but is more about the growth of consumption options, which now includes near-live content such as sports or news, distributed via peer-to-peer networks. In some respects the peer-to-peer sites have been test-beds for the content industry, leading, for example, to BitTorrent’s ‘tit-for-tat’ protocol that transfers video files in parallel, sharing the load among multiple computers. The protocol is so called because it rewards users for allowing their hard drives to participate in file distribution by in turn being able to exploit more of the network’s peers for downloads.

Peer-to-peer

The peer-to-peer method dissociates the video download (or streaming) rate from the performance of the peers, and exploits the previously under-utilised uplinks within the access networks. But in terms of bandwidth as a whole, the protocol achieves nothing, as some operators found to their cost. The UK’s cable operator NTL found that BitTorrent worked almost too well for its video download service, clogging up its core bandwidth. The remedy was to implement edge servers from CacheLogic to store more popular content for delivery directly to users without peer-to-peer traffic consuming core bandwidth.

The NTL case highlights the kinds of compromises that will be needed for IPTV in general, mixing storage and forwarding technologies to suit different situations as the variety and requirements of content broadens, ranging from mobile clips to HD movies, niche or local content to highly popular set piece broadcasts like major sporting events.

COD challenges

The P2P experience has exposed some other critical issues as well. We can say that the Content-on-Demand (COD) revolution brings challenges under four headings: bandwidth, contention, protection and aggregation. Bandwidth we have already considered; the problem being that while P2P will have its part to play for the so called ‘long tail’, it does not scale well for popular programming, where the requirement is for rapid concurrent access by many users. This raises the question of what kind of storage to have and where to locate it within the overall system architecture. Cache memory is required alongside disk drives within the storage hierarchy to provide the level of throughput necessary to support high levels of concurrent access to content. The growing viability of flash memory for video caching, rather than the longer established option of DRAM, is an important development with two advantages: lower cost and – being non-volatile – immunity from power failure, which in turn creates further economies by requiring less sophisticated power management.

Until now, flash has been held back on two counts: by being less dense, and only supporting a limited number of memory over-writes, which is obviously not much good for ever-changing video. These problems, though, are well on the way to being solved, offering service providers a highly effective new option for caching. For this reason, flash memory is being hotly pursued by several of the leading vendors of video servers, notably C-COR.

But this begs another question: where to put the cache. The conventional wisdom, stoked by some of the server and storage vendors because this helps them sell more equipment, has been towards replicating popular content and pushing it out towards the edge of the network in order to save bandwidth in the core. But the growth of on-demand viewing is causing some operators to reconsider and even reverse such strategies, with centralised approaches regaining their former appeal. This is largely because the task of managing a distributed approach becomes increasingly complex as the content mountain accumulates with ever more concurrent access. Furthermore, cache – even if based on flash memory – will remain a precious commodity that will be easier to utilise efficiently if it is all in one place and can be marshalled by a coherent global file system. At the same time, network bandwidth continues to become cheaper and more plentiful with the help of optical transmission technologies such as DWDM.

The P2P sites have also provided valuable pointers concerning the other two issues, aggregation and protection. The story of Google’s acquisition of YouTube has been widely told, highlighting as it did the failure of Google Video to gain traction largely through its lack of aggregation, which offered little help in locating desired content. This was one of YouTube’s strengths, however much one might find low-resolution content to be almost unwatchable on a large screen.

The message is that intelligent aggregation and strong video search capability will be important competitive factors for IPTV services. Indeed, as more content becomes available on demand, consumers will want ever-greater help in sifting through it. Increasing freedom from rigid schedules could, paradoxically, create new demand for personalised channels that present viewers with content tailored to their own tastes.

Yet it is over the thorny ground of content protection where the P2P companies have experienced the greatest teething pain, although it could be said that the studios and producers themselves are sharing the burden as the whole broadcast and film industry evolves towards a new model for rights management.

YouTube has been at the centre of the debate and exemplifies the rather unholy position the video sharing sites have held of owing their success to the posting of unauthorised material. The company has usually removed such material when asked to, and has cooperated with the rights holders in prosecutions by providing user identification information, but has attempted to deflect responsibility for violations to the perpetrators. Such aversions have enjoyed mixed fortunes under the light of legal scrutiny and, in fairness to YouTube, the company has in practice recognised that it has a part to play in defending intellectual property.

Indeed, the P2P sites realise that their ability to combat unauthorised distribution of content must be part of their competitive strategy for dealing with the big houses. For example, YouTube’s own deal with CBS allowing users to access news, sports and prime time programming includes technology that allows CBS to detect and remove unauthorised content. Furthermore, the advertising model is gathering steam as a way of deriving money from content, perhaps even when the original posting was not authorised. So we note that CBS also has the alternative option under its arrangement with YouTube of accepting the posting and making money instead by streaming advertising with it. Meanwhile Google has signed separate deals with Sony BMG and Warner Music to offer music content supported by advert streaming, although in Warner’s case there is a purchase-for-download option.

Yet despite this progress, the content protection issue still has some way to run, with sites such as the Chinese owned TVUPlayer facing the same kind of action as Napster in the early days of peer-to-peer music downloads. The difference with video, though, is the substantial role that advertising has in oiling the transition towards a legal download model.

For the sites themselves, there are other money making avenues opening up, with emerging applications such as Telepresence looming in the wings – but that is another story. There will also be a growing market for sources of high-definition content as the population of HD ready screens continues to grow.

First published April 2007 in IPTV News Analyst (see http://www.digitalmediapublishing.co.uk)