Most countries have taken steps to force broadband service providers to give much more transparent and realistic QoS (Quality of Service) indicators, so that consumers can compare individual packages on quality as well as price. But there are increasing moves to go one step further by enforcing minimum QoS standards, which would be reviewed periodically as technology improves and expectations increase. This is closely allied to the network neutrality issue, which has reared its head again in some countries, especially the US, where Comcast and Time Warner have been investigated by the Department of Justice, accused of throttling OTT video services that compete with their own, notably Netflix.
The net neutrality issue has been particularly tied in with moves to legislate for QoS in France, where communications regulator ARCEP has hinted it will go further than just monitoring QoS and set minimum QoS requirements that operators will have to meet to retain their licenses. These plans have been outlined in the regulator’s latest report on network neutrality, which were submitted to the French government last week.
For this report ARCEP has also taken an inventory of traffic management practices implemented by operators, and has not liked what it has found. The regulator has stated that some of the practices, such as blocking IP voice services, and also peer-to-peer traffic on mobile networks, run completely counter to the spirit of its framework on network neutrality set out two years ago, which was based on five criteria. ARCEP is referring here to the differentiated treatment of streams via throttling, blocking or priority queuing. It insists that such practices be eliminated, and if this does not happen voluntarily will invoke existing legislation to enforce it.
Monitoring of network neutrality will then go hand in hand with QoS and be performed continuously, while also probing the performance and operation of interconnections between internet service providers. However ARCEP is concerned that imposing rules on interconnection ties and their pricing could create friction between the players, and so does not at this stage plan to include these within its regulatory framework, although it will monitor these much more closely.
The network neutrality debate in the past has focused largely on unfair competition, and abuse of power by the big ISPs, which have been accused of exploiting their position in broadband provision to favor their other services, such as voice over IP, and especially pay TV. But the debate is now exposing other issues involving the distinction between different types of service and content types. Firstly it is becoming clear that if bandwidth is limited it makes no sense to treat video, conferencing, voice over IP, and data services such as email all the same. They have different requirements for bandwidth, latency and jitter, and these should be taken account of in traffic management to optimize each application or service separately. Secondly mobile and fixed networks have different dynamics that cannot be readily addressed via a single one size fits all approach to network neutrality and QoS regulation.
Taking the second of these issues, consider Brazil, where the regulator Anatel has already enacted rules over QoS. The problem is that Anatel has set the same rules for fixed and mobile operators, with similar minimum performance requirements for each.
This is understandable in the sense that large parts of the country rely on mobile communications for broadband service, just as in the past many people in more remote areas have received TV over MMDS radio networks that are only now being phased out. By February 2012, there were 45.2m mobile broadband connections in Brazil, almost three times the 16.5m broadband connections. Given that fixed broadband is unlikely to reach vast swathes of the population anytime soon, Anatel decided it needed a common regulatory approach to both, but has made the mistake of being too rigid here. While not actually enforcing the same specific speeds, it is requiring both fixed and mobile bit rates to rise year by year at the same rate, imposing an expectation of improvement as technology evolves, and as fiber goes deeper.
The rules are quite specific, taking as a baseline the maximum speed set out in the contract, and then requiring the minimum connection rate during peak traffic times never to fall below 60% of that for the year ending August 2013. This will increase to 70% the following year and 80% for the year ending August 2015. This seems a ludicrously tight band, although it will have the effect of ensuring operators become very careful to avoid advertising unrealistic headline rates. But it fails to recognize that over mobile networks, bit rates are subject to much greater fluctuation according to congestion within cells, and also environmental factors. The effect of the new rules could be to limit the general QoS available over emerging 4G networks in Brazil.
The other issue relates to the content itself. Streaming video can tolerate considerable variations in bit rate, but not in jitter. Interactive video and voice is particularly intolerant of high end to end latency. Other forms of data require guaranteed delivery, but can stand considerable delay. These differences have traditionally been accounted for more at the application level, but that itself can impose delay and there has been increasing interest in media aware networks that do take note of the differing QoS requirements of traffic. This is not against the spirit of net neutrality providing there is no prejudice against particular services or applications for commercial or political reasons, only on pure QoS needs. For example T-Mobile’s blocking of Skype services is strictly non neutral, and even worse was Comcast’s obstruction of forums critical of its business practices, which is not so different from certain despotic regimes.
On the other hand there are some cases where discrimination on an application or source basis is in the interests of many users, as with data intensive file-sharing platforms that hog bandwidth, often for illegal content. The net neutrality moral high ground blurs here, since it can be argued that all users should have equal rights of access on a data volume basis. The scope for debate and disagreement here will never go away until there is a balance between core and access bandwidth that really reflects the Internet’s emerging status as a universal platform for always on real time multimedia services, rather than sporadic transmission of best effort traffic such as email. To some extent this transition is being driven by the proliferation of CDNs (Content Delivery Networks), which in effect allow accelerated long distance transport of video services, which can then be plugged into the access networks of the users, bypassing the public Internet substantially or altogether.
Clearly the Internet is in effect fragmenting into a multi-tier delivery infrastructure, and there is no going back on that. Differentiated QoS is here to stay, but there is a possible compromise that recognizes the clear distinction between best effort and real time traffic. We do not have to look too far to see this model already in action, in the German Interstate Treaty on Broadcasting and Telemedia (called Rundfunkstaatsvertrag). This ordains that only one third of a platform capacity is totally under the operator’s control for dedicated real time services, with the rest being reserved for best-effort traffic. Furthermore the QoS outside the special lanes and premium services paid for by other service providers must not fall below a guaranteed minimum.
Of course this does not extend to external CDNs, and only applies to broadcast infrastructure, but the principles are being enshrined in the recently submitted ministerial draft bill for the amendment of the German Telecommunications Act (Telekommunikationsgesetz ‘ TKG). This follows the line of recent European Union directives in giving the German Federal Network Agency (Bundesnetzagentur ‘ BNetzA) the potential to define minimum requirements for service quality. The aim is to ring fence some capacity outside premium services, conceding that not all data traffic is equal, but insisting that it should all have certain minimum rights.
This draft bill has actually disappointed advocates of strict network neutrality, but appears to be the way Europe is heading at the moment towards a classic fudge that does at least enshrine some basic rules for best effort services, favoring industry self-regulation rather than government intervention. However France at least has indicated it is already losing patience with the voluntary approach.
This analysis was first published in Wireless Watch’s sister title Faultline, which provides weekly research notes and analysis on the worlds of digital media and quad play.