< Back

EU net neutrality regulation must evaluate QoS practices

Mon, 8 Jul 2013


Net neutrality issues cover two technically similar but commercially distinct areas. The first is the issue of competition, which means ISPs restricting traffic to and from applications that compete with their own product range. Two clear examples are the texting application, WhatsApp and the VoIP system, Skype.

Both services challenge traditional texting and voice communication (meaning 2G cellular or PSTN) services of the telecoms companies upon whose networks these applications rely. By degrading or blocking these new services out right, ISPs are accused of restricting application innovation and a free and open market.

WhatsApp and Skype challenge the traditional texting and voice communication telecoms services upon whose networks they rely.

The other part of the net neutrality argument concerns what’s called throttling - there are a number of other descriptions: packet preferencing, traffic shaping, traffic engineering and so on. Here, ISPs again preference or de-preference certain traffic depending on its source / destination, this time with the intention of creating a two speed internet, one where a faster and more reliable service can be offered to certain kinds of traffic, for a fee.

This traffic is generally associated with bandwidth hungry media providers such as the BBC’s IPlayer or other major traffic sources such as Google, Amazon and Facebook. The point here is that by instituting this kind of traffic engineering and their resulting revenue streams, ISPs are are again accused of mis-selling their service, a service that states a certain Mbps with the inherent understanding that traffic discrimination will not be an issue.

Traffic from larger web companies, especially content providing organisation, is vulnerable to traffic enginnering

Technology surrounding this traffic engineering capacity is fairly well known, at least in the networking industry. In short, by simply inserting certain flags and markers on packets from or to certain IP destinations as they enter the network, a traffic shaping policy can be maintained.

Furthermore, with the advent of newer technologies such as MPLS, this shaping - or Quality of Service, QoS - capacity has been enhanced allowing content types such as streaming media or VoIP traffic to be more easily manipulated within a network.

With packet preferencing becoming easier to manage through such services as Cisco’s Content Deliver System, ISPs, by having restrictions on their network, are in a position to increase revenue by demanding that media or VoIP providers - or end users - pay for an enhanced service.

ISPs are in a position to increase revenue by demanding that end users, either households or media giants, pay for an enhanced service.

The argument for and against net neutrality has meandered on for a number of years. Those demanding a fully even playing field with no QoS activity include luminaries such as Tim Berners-Lee and political movements such as the Pirate Party. Their argument covers both economic and human rights platforms (including freedom of information), stating that by creating arbitrary restrictions at the IP level, ISPs are exploiting their technology to the detriment of the free market, innovation and general social advancement. They argue that a law is needed to curb these activities.

Internet inventor and luminary, Tim Berners-Lee

On the other hand, whilst many ISPs agree with the free market argument, they argue that throttling services such as BBC IPlayer, Google and Amazon traffic at peak times is necessary so that all users receive an equal service. However, the truth is that these restrictions are by no means inevitable. In fact, it could be argued that the perspective on throttling offered by ISPs is somewhat misleading. The point being that any discussion of traffic shaping starts with the basic premise that capacity on the network is restricted, meaning they network itself is over capacity.

In which case, far from defending their customers rights to a certain level of service (the majority of those customers who in any case have chosen to access a particular media source), ISPs may be accused of throwing a smoke-screen over their own lack of infrastructure investment.

ISPs may be accused of throwing a smoke-screen over their own lack of infrastructure investment.

Interestingly, this point was mentioned explicitly in guidelines published as far back as 2008 by a Japanese working group set up to ponder issues of net neutrality, which of course exists in Japan as much as any other internet orientated country. Packet shaping is not an excuse to avoid investment, their paper pointed out; ‘ISPs should increase network capacity when network traffic increases’.

Individual nations have been grappling with the issue with different results. The Netherlands and Chile have enshrined net neutrality into law. In the US, the Federal Communications Commission (FCC) has also determined regulations, although loopholes are being readily exploited and the existence of the regulations themselves is presently up for debate – see this article. The European Union has dragged its heels.

However, as recently as June 4th, in a statement by Neelie Kroes representing the Digital Agenda at the European Commission, some indication of how future proposals for regulation were provided. Competitive blocking (it seems) will be ruled out but traffic engineering, partly in line with the Japanese position, in principle will be tolerated to enable ISP to gain revenue for further investment.

Kroes says, ‘The fact is, the online data explosion means networks are getting congested. Internet Service Providers (ISPs) need to invest in network capacity to meet rising demand: and the right predictable regulatory framework will help them do so. But, at peak times, traffic management will continue to play a role: and indeed it can be for legitimate and objective reasons; like separating time-critical traffic from the less urgent.’

Kroes is on record for defending the rights ISPs have to offer different kinds of connections to consumers - cited in this article. All the same, what is not clear is whether ISPs are free to open further revenue streams by levying charges on content providers themselves – a position endorsed by UK Culture Minister Ed Vaizey in 2010 in a speech at the World Telecoms Conference - see this article.

UK Culture Minister Ed Vaizey and Neelie Kroes, Digital Agenda at the European Commission

So what does this mean for data centre clients with London colocation or other services that take internet feeds from ConnetU? Firstly, the ConnetU network imposes no QoS or traffic engineering whatsoever and has no need to do so as it maintains a large surplus bandwidth capacity. The only options are the size of the pipe and/or the possibility of a lower cost partial feed, which makes sense if a significant amount of a customers traffic can be passed over an internet exchange – see this blog for more on internet exchanges.

The real problem for customers of any colocation provider is that, depending on which network their traffic traverses, they may be subject to the same restrictions as everyone else once their packets hit a network that has decided to regulate certain types of media.

Content providers, for example, may find that their services are degraded once packets have departed the ConnetU network as the end user may be connecting via one of these restrictive ISPs. This is, of course, unavoidable as it is impossible to dictate which ISP a consumer or business will use.

Content providers may find that transit services are degraded once packets have departed the ConnetU network.

In the UK, a number of well known ISPs have signed up to a voluntary code of conduct put together by the Broadband Stakeholders Group known as The Open Internet Code of Practice. The code stipulates that signatories must make known the methods by which they throttle traffic at peak times.

The code therefore follows the more liberal aspects of the EU Commission position, specifically in its Commitment 2 which states that ‘ISPs retain the ability to deploy reasonable traffic management practices over their networks’.

It goes on to list those practices, the first of which is ‘managing congestion on its network’, a handy catch-all position. Though many UK ISPs have signed up, there are yet a number who refuse as even these very broad and undefined practices are deemed too restrictive – see this article for a list of signatories and non-signatories.

The position of the major ISPs would appear to be: let us charge more and we will be able to invest in our infrastructure. At the technical level, this should ultimately lead to a situation where, like ConnetU, there is simply no need to employ traffic engineering as there is plenty of available bandwidth to cover all demands.

Yet the question remains at the commercial level: is it reasonable to expect a business to simply close off a profitable revenue stream once in has been opened and instated in the minds of its customers?

Proper investment should ultimately lead to a situation where there is simply no need to employ traffic engineering.

Furthermore, how are we to regulate investments made with these extra revenues by ISPs. It must be clearly understood that QoS, traffic engineering, is a way to make networks more efficient when there is a limitation on bandwidth – but with new transmission technologies and ever more powerful and cheaper hardware, this limitation is not, or should not become, an acceptable status quo.

The risk is that, with the belief that they are promoting a necessary open market for ISP traffic delivery, the authorities merely succeed in sanctioning underinvestment in critical networking infrastructure by certain ISPs.

A market for preferential packet delivery could simply lead to ISPs eschewing newer, faster technologies in favour of opening up retrogressive revenue streams that will only serve to keep UK internet infrastructure in the dark ages. The question is, will new EU regulations take this risk into account?