According to the new Ericsson Mobility report, video traffic is growing by 60% a year and will be driven further by rising prevalence of fast LTE networks. Video consumption is on average 2.6Gbytes per subscription per month in some networks, says the vendor, and best effort is no longer good enough for many consumers ‘ faster networks are leading to higher expectations of user experience, and Ericsson detects a close link between network quality and subscriber loyalty.
All this means that the Mobile Video Global Summit, held last week in Germany, was particularly timely, since carriers are increasingly preoccupied with the difficult balance between delivering a strong video experience and reining in network capex budgets. The timing proved even better, since the European Commissioner for digital economy, Neelie Kroes, chose the same week to reiterate her stance on net neutrality ‘ comparing networks to roads, she said that, in order to avoid anarchy, roads must have traffic lights, speed limits and a highway code, and they will not necessarily be free of charge.
This was the backdrop to an event which was primarily focused on optimization, with DPI (deep packet inspection) vendors like Allot Communications, Quickfire Networks, Avvasi, Vantrix and Saguna there alongside cloud players like Equinix and the constant reassuring presence of Cisco.
Their overall message was that ‘we are just trying to help the cellular operators’ with better video tools and analytics, such as reporting techniques that indicate when an OTT video was stalled, or whether a consumer prefers a stalled video which then buffers and plays in HD, or an SD experience which plays now.
There was a broad assumption that video on mobile networks is over-the-top video, but the elephant in the room was very quiet and came in the form of operators such as UAE-based Du, Denmark’s TDC and Deutsche Telekom (to name a few), who have their own HD video services, mostly IPTV, that can also be also found on their mobile networks.
Do they favor their own content? In effect they do, though in slightly different ways which may prove relevant to the neutrality debate. TDC has no plans to introduce seamless Wi-Fi roaming in its territory, because ‘consumers all know that it’s cheaper on Wi-Fi and switch themselves’ ‘ in most cases to Wi-Fi provided by TDC. Berdn Reul, VP of product management, shrugged at the idea of seamless roaming. It was not his problem.
Des O’Connor, VP of data design at Deutsche Telekom, was a little more vague. He conceded that DT ‘did what it could’ to ensure that its T-Entertain content was delivered in the best shape it could, without wishing to go into detail and while assuring us that it was all done within the guidelines of Neelie Kroes.
Despite these operator efforts, the DPI firms mainly focused on the rise of YouTube. As highlighted by Ericsson, this is as much down to the switch from 3G to LTE – Ben Cox, VP at UK lead cellco EE, pointed out that this increased traffic by 2-3 times simply because YouTube would increase to HD when running on an LTE network.
In this kind of company, there is a real sense that cellular operators are the passive bystanders who have really been wronged. YouTube eats all the bandwidth that it can and the best thing that DPI solutions can do is to recognize it and force it to drop to a slower bit rate, rather than step up to the next level. As Marc Zionts, VP of strategic mobile service providers at Allot, said: ‘It was a protocol written by handset players and content people, to try to grab as much bandwidth as it can, whether it needs it or not. No cellular people were consulted when it was written.’
So Kroes will have to tread carefully in the formulation of her neutrality principles or all of the auctions for LTE spectrum could have been held in vain, and new networks will produce little more than millions of dissatisfied cellular customers who cannot even see cheap and nasty YouTube videos, because their protocols demand the type of bandwidth that should be reserved for UltraHD soccer games.
Gail Smith, partner at Cavell Group, talked about how the cellular world is fighting back. After a variety of references to how badly YouTube plays on the Free cellular network in France, she added: ‘It plays pretty badly on every ISP in France. Orange claims to have got money out of Google for delivery into its African networks, but this was more to do with ISP peering than net neutrality, and that’s Kroes’ Achilles heel. Like Comcast in the US, which charged the Netflix ISP more for what had become a heavily imbalanced peering arrangement, France Telecom and Free and others have attacked by either increasing their peering charges to the company that YouTube uses, or by failing to increase its throughput unless more funds were made available for peering. It shows that there is more than one way to skin the net neutrality cat.’
But while YouTube wilts on the mobile in France, the result simply infuriates worldly French consumers, who blame their cellular suppliers instead of blaming YouTube.
If it weren’t for the fact that cellcos throw off so much cash right now, despite the threat of declining voice revenues, the upgrade costs to LTE, the high price of spectrum, the rising tide of video data requirements, and the indifference of both handset vendors and content owners to their plight, you might feel sorry for them. But in the midst of a global recession and a weak European market, the thoughts of Kroes are with the consumer.
There was one dissenting voice among all those nodding heads which embraced traffic shaping and video optimization. Louis Plissonneau, 3G IP traffic analyst at Orange France, pointed to some research he had commissioned on YouTube. He showed a graph of data usage when an iPhone downloads a single clip of YouTube and another when a Samsung S3 downloads the same clip. The Samsung uses half as much data, because it adopts an approach of bursting four seconds of download at high speed and then playing it, while the iPhone drains data at its own rate. Plissonneau pointed out that, if the user stopped the download on the iPhone, it had already collected a great deal of data in advance. Given that many people stop viewing and therefore stop the download, but most of the video had been delivered anyway, this was wasteful. It also shows how Apple can impose protocols and data usage patterns on operators with an unhealthy disregard to the effects on the network. For now at least, operators have less power to pressurize Apple to fix its problems than other device suppliers, though this may well be set to change.
Kroes is adamant that she will introduce legislation on net neutrality, albeit carefully and with a heavy heart. She seems happy with traffic management, where such a policy is purely for the health of the network, not for financial gain, to protect time critical traffic from the less urgent traffic. Where she will hit hardest with regulations will be where online services are blocked or degraded, often without consumers’ knowledge ‘ a recent European study said that one in five fixed lines, and more than one in three mobile connections, experience this. Start-ups are being targeted because they have less weight to fight against operators, says Kroes.
Her other big issue is that many Europeans are not getting the broadband speed they are paying for or which they see advertised. In this area, the technologists at the conference were calling for standards and promising that in 2-3 years’ time, they would have a handle on what causes video to stall, and would be able to tune their networks in real time. But even then they may have no idea of what speed they could advertise, because the real speed depends on the nature of the traffic, what the protocol is doing, and who else is on the network.
The Kroes comments brought immediate frustration and reaction from leaders at the GSMA, who expressed ‘disappointment’ at the idea of early intervention on net neutrality. However, many of her hot targets are scarcely relevant to cellcos. For instance, she aims to come down heavily on operators that want to block Skype or WhatsApp. Most cellcos seem to be barely at the point of tracking these apps and working out how much traffic they generate, and just on the cusp of understanding how to smooth their service delivery. So Kroes’ actions may be more about preventing future bad behavior, as she sees it.
And attacking YouTube peering partners is not the only way to improve the economic balance of mobile video. In the final panel debate of the Berlin show, DT’s O’Connor was asked whether the right approach to video might not be to partner with key players over a managed part of the network. He said: ‘That’s the entire reason I am at this event,’ and talked about creating and exposing network APIs to such partners. An earlier presentation from EE’s network strategy manager Matt Stagg had isolated a magic number, which was ‘how much capacity do you think you could harness for privileged partners to make sure that their video hardly ever stalled and which enjoyed low latency?’ The answer, everyone seemed to agree, was 15%, and then the rest of the traffic would not get anything other than best effort ‘ but that would not be noticeably worse best effort than it is right now.
In other words, operators would charge some kind of premium for services up to 15% of their total data capacity – whether that was paid for by revenue share, payment, shared risk or other options.
O’Connor was also impressed by what Philippe Brodeur had achieved at AerTV, the over-the-top subsidiary of Magnet in Ireland, which had put together an IPTV service, and had also added some extremely low cost channels of its own, one in particular that focused on regionalized and local sports. These are the kind of initiatives we can expect to see from larger operators ‘ while the walled garden may be dead and true video is OTT, it can still be better managed if it is in the data portion of the delivery, not necessarily lumped in with all the best effort internet.
When faced with the idea that TV Everywhere strategies will soon make up the greatest portion of video to a portable device, the entire room was confident that YouTube was the real worry, since that is consumed on the move, and so goes over cellular. Wi-Fi has carried the brunt of TV Everywhere so far, both in Europe and in the US, with figures as high as 80% or more of such video being watched on Wi-Fi without any formal offloading other than the consumer choosing a cheaper connection. But truly mobile consumption could change that balance. There was a general feeling that content providers, and Kroes, needed a better understanding of how much mobile carriers were investing in their networks, and how much over-the-top players actually get from cellcos in terms of quality of delivery. Operators are so busy trying to keep up with the requirements of cellular data these days, they often forget their mantra, ‘we are more than just a bitipe’. Without buying broadcasters, contracting with content, speeding up TV Everywhere and rebuilding the garden walls, they risk becoming little more than that bitpipe, which in a few years’ time, really won’t be throwing off quite so much cash.
This analysis was written by Peter White of Rethink Technology Research, who chaired part of the conference and heads up the Faultline research service. This dissects and comments on technology and business aspects of the digital video market on a weekly basis.