The net neutrality service model explained
by Frode Sorensen
This year, the net neutrality debate has been raised to a higher level. It has not gone unnoticed that the European Commission has proposed a new Regulation on European single market for electronic communications which may mean “ratification” of net neutrality. The political process is moving ahead at full speed to handle the proposal through to its final destiny.
Originally posted at Norwegian Post and Telecommunications Authority.
Also published in “Network Neutrality: an Ongoing Regulatory Debate“, 2014 Annual Report of the Dynamic Coalition, edited by Luca Belli & Primavera De Filippi, preface by Vinton G. Cerf
Furthermore, the Model Framework on Net Neutrality has been launched subsequent to the European Council organizing its multi-stakeholder dialogue in May. The framework has been established by the Dynamic Coalition on Net neutrality, and the launch itself was made on the Internet Governance Forum in October. The purpose of the framework is to make recommendations to lawmakers about how best to ensure net neutrality.
In many ways, these two events draw “conclusions” regarding the previous years’ development in the field. The U.S. regulatory authority, FCC, was the first to present its principles regarding the open Internet in 2005. The debate came to Europe shortly after, and discussion on the topic was high on the political agenda during the revision of the Telecom package which was eventually adopted toward the end of 2009.
Meanwhile, the debate was also significant at the national level, and in some countries this resulted in the establishment of national arrangements. In Norway, early in 2009, the Norwegian Post and Telecommunications Authority (NPT) took a co-regulatory approach in order to bring the industry together on agreed national guidelines for net neutrality. And in 2010, BEREC Expert Working Group on Net Neutrality was established under the chairmanship of NPT.
Net neutrality is the principle that all Internet communications shall be treated equally. Equal treatment of traffic means that the traffic is transmitted irrespective of content, application, service, device, and irrespective of the sender or receiver. The latter element means that transmissions shall be carried out equally for different end-users, including content and application providers.
Up to the present moment it seems that the understanding of what net neutrality is and what it is not has matured significantly. A consensus was established relatively early that net neutrality applies to the Internet, and not to other forms of electronic communications networks. Essential concepts in this context are so-called “specialized services”, i.e. services that are not Internet access services.
The Norwegian guidelines for net neutrality stipulate that “if the physical connection is shared with other services, it must be stated clearly how the capacity is shared between Internet traffic and the other services.” When the guidelines were established in February 2009, specialized services were not a familiar concept, and therefore this term was not explicitly used. The FCC has not defined specialized services in detail, but writes, for example, about “specialized services, such as existing facilities-based VoIP” in its Report and Order in December 2010.
In 2011, BEREC established this definition of specialized services: “Specialized services are electronic communications services that are provided and operated within closed electronic communications networks using the Internet Protocol. These networks rely on strict admission control and they are often optimized for specific applications based on extensive use of traffic management in order to ensure adequate service characteristics.”
And in the framework of net neutrality, the Dynamic Coalition on Net Neutrality defines specialized services as “electronic communications services that are provided and operated within closed electronic communications networks using the Internet Protocol, but not being part of the Internet. The expression ‘closed electronic communications networks’ refers to networks that rely on strict admission control.”
Since the whole idea underpinning net neutrality is to ensure equal treatment of traffic, and specialized services are exempted from net neutrality considerations, it is essential that the specialized services do not have a negative impact on Internet traffic. Otherwise, this would effectively undermine the foundation of net neutrality. For how will mutually neutral handling of traffic help if external conditions degrade the capacity available for Internet traffic as a whole?
Specialized services can help satisfy the need to guarantee the quality of certain forms of communication. As the definition of BEREC stresses, such services could be optimized for specific purposes. A typical example is real-time services such as telephony and the like. Specialized services can be provided with support for quality of service by having the services set up in networks where capacity is dimensioned in relation to amount of traffic, and the traffic load is made predictable based on access control (typically based on a subscriptions).
Quality of service to specialized services is not secured by giving these services an explicit higher priority level than the Internet access service, but rather by having adequate capacity reserved for the specialized services without this being done at the expense of Internet traffic. Internet traffic has its own capacity scaled according to the contractual access speed. (The latter should not be understood as the Internet access service has an absolute guarantee relating to speed, but this is, however, based on statistical calculations.)
The importance of separate capacity for the two service categories is also very evident in the BEREC definition of specialized services. These services are offered in “closed networks” that make it possible to separate this traffic from Internet traffic. Both service categories are typically transmitted over the same physical infrastructure, in which case, sufficient resources are to be available for the specialized services and Internet access service in their own “logical network”.
Such “closed networks” can help to ensure that specialized services do not have a negative impact on the Internet access service, nor degrade it. This is already clear from the current Telecom package: “In order to prevent the degradation of service and the hindering or slowing down of traffic over networks, Member States shall ensure that national regulatory authorities are able to set minimum quality of service requirements on an undertaking or undertakings providing public communications networks”. (USD 22.3)
The devil is in the details
Having said this, it is also important to emphasize that there is nothing negative in traffic management in itself. Traffic management is called for if one is to efficiently handle the traffic in the networks. In connection with net neutrality, a distinction is made between reasonable and unreasonable traffic management. Unreasonable traffic management is basically traffic management that provides non-neutral transmission of different types of traffic. But exceptions are also made in individual cases, which may in some cases be regarded as reasonable.
BEREC has defined four criteria for reasonable traffic management: non-discrimination of content and application providers, end-user control, application agnosticism and proportionality. BEREC also emphasizes that these criteria should not only apply to technically implemented traffic management, but also to other restrictions such as, for example, described in contractual terms.
Typical exceptions which may be considered reasonable are: (1) orders given in statutory bodies of law and court decisions, (2) measures to ensure the integrity and security of the network, (3) the prevention of unsolicited communication, (4) measures based on a direct request from the end user and (5) dealing with special situations relating to congestion management.
Most of these exceptions are easy to understand. Net neutrality should not be used to legitimize illegal or harmful activities (items 1 and 2). The problem of spam and the like must be handled efficiently (item 3) and the end users must be able to protect themselves on their own access when this does not affect others, such as parental control (item 4).
The exception that is the most complex is how to deal with congestion. The way the Internet is constructed means that congestion will necessarily arise from time to time. Internet service providers’ main measure to deal with this is to build capacity in the network in accordance with the subscription agreements entered into. Moreover, short-term congestion will automatically be handled by the built-in mechanisms in IP technology.
If there is a need to manage the traffic load above and beyond this, the mechanisms that handle the various applications neutrally (application-agnosticism) and allow an end user to decide what his available capacity will be used for (end-user control) should be preferred. Only in special situations where this is not possible in practice, should it be necessary to make use of application-specific methods.
Toward pan-European net neutrality?
It is a positive development that the situation is moving from a fragmented approach in various countries toward a common European approach to net neutrality. The service model consisting of the two categories of services, Internet access service and specialized services, is an important foundation for future efforts to unite on a common understanding of net neutrality. This methodology also appears to be established at the global level today.
However, there is still a need to clarify the understanding the model:
- The model assumes that the two service categories are defined as clearly as possible, so that there is no “playing” regarding which label you put on the service provided.
- Further, if the model is to work, it is essential that resources to the service categories are separate (“closed networks”) to avoid degradation of the Internet access service.
- And finally, management of “traffic jam” (congestion) is by and large done irrespective of the applications, and only in special situations where this is not possible, may it be application-specific.