Personal tools
You are here: Home InternetNZ Activity Submissions Archive 2006 Submission on the Draft Paper: “Internet Interconnection and Peering Report”

Submission on the Draft Paper: “Internet Interconnection and Peering Report”

InternetNZ welcomes the opportunity to provide comment on the draft paper “Internet Interconnection and Peering Report”, provided in a draft format for preliminary comment.

InternetNZ acts as an advocate for an open and uncapturable Internet, with an interest in ensuring that developments within the Internet generally are not able to unreasonably advantage specific sectors. Peering is one such component of the Internet's infrastructure and technical history that requires care and concerned interest in order that the Internet can continue to operate effectively and fairly for users.

In a general sense, InternetNZ believes that the report as provided provides an over-emphasis on descriptions of network connectivity and networking, with too little consideration of the factors that promote good technical engineering design. Peering is an advantage to the users of the Internet, and it provides network reliability overall.

The paper seems to imply that the current situation in New Zealand must be the right one, because it is the one that has developed by allowing market forces to prevail. Good networks do not develop in that manner, and InternetNZ believes that encouragement and even direction may be required to ensure that the Internet continues to provide an opportunity for innovation and the best possible user experience.

InternetNZ believes that the paper does not provide an appropriate description of the range of peering behaviours that can be developed between ISPs, content providers, transit providers and organisations that might provide more than one of these functions.

W.B Norton (Equinox) in his range of White Papers on peering describes the range from no peering, to open peering, to selective peering, to restricted peering. In each case, the decision-making processes involved (technical, economic, philosophical, marketing) may lead to specific differences that are not well covered in the paper. The term 'peering' can be a very blunt description of a whole range of agreements that vary considerably in scope and application.

Consider the case of a small regional ISP who wishes to develop a peering relationship with a larger national ISP. It would be entirely wrong to assume that the smaller ISP would invariably receive undue advantage – it all would depend on the peering policies that are negotiated. In most such arrangements, the movement of traffic is restricted entirely to the two networks – the smaller ISP would not be able to use the arrangement to unfairly use the national network of the larger carrier to move traffic to another part of the country to be delivered into a third provider's network. That traffic would still have to travel through the smaller ISPs existing transit arrangements. The peering would only be available for data that originates in one of the two networks and terminates in the other.

If the larger ISP was concerned that the smaller ISP, lacking its own national network, was gaining too much advantage by 'piggy backing' traffic to the other regions of the larger ISPs customer base, the arrangement could be restricted to the exchange of traffic only intended for the local, originating region. The paper refers to the problems of 'backbone freeriding' and suggests that the only fair solution is transit. While that might well be one solution, the smaller ISP might also wish to use other transit arrangements to get data to the other national locations – to imply that a paid transit is the only fair outcome is not valid.

The point is that peering will almost always provide some degree of advantage to both parties, and the variations to peering policies should be used to fine tune the issues of fairness, rather than allowing it to become an issue of economic manipulation.

Traffic from a customer to an ISP (and therefore the Internet as a whole) is effectively paid for by the customer, and that customer should have a confidence that the data will delivered in the most efficient and expedient way.

If ISP "A" will only accept traffic if paid by ISP "B", the traffic is in a sense paid for twice by ISP "B", and giving ISP "A" an economic advantage. In the long term it will work to make the smaller ISPs uncompetitive, and work against diversity of opportunity for end users.

Peering has always been encouraged from a technical point of view because of the diversity of routing that it provides. This diversity is at the heart of the Internet's success to date, and InternetNZ believes
that it should be fostered and encouraged. Diversity of pathways provides for increased reliability and stability through the diversity of the interconnecting networks that comprise the Internet.

End users are always advantaged by rich and complex peering arrangements. The user experience – the perceptions of speed and throughput – are enhanced by the expedient and efficient routing of data through the network. Any arrangements that require data to flow in a less than optimal route mitigate against the user experience.

Peering provides for better network management generally, with the opportunity to identify and route around problems relating to outages and congestion. Any network that limits the number of points of connectivity become prone to outages. The data outage in mid-2005 is a prime example of a network that lacked diversity – the combined efforts of one digger and one rat were able to interrupt telecommunications services for more than 100,000 consumers. Such a thing should not be
allowed to happen, and certainly not excused by any claim that it was simply a market/economic decision for the provider. New Zealand's critical infrastructure deserves more care and planning than that.

It is a nonsense – both technically and economically – for data that originates and terminates in New Zealand to ever be required to transit overseas. While the paper would suggest this is rare (referred to as a 'claim' or 'misconception' in the paper) it is the inevitable outcome from a lack of peering within NZ by NZ ISPs and content providers, and it has a detrimental impact on the Internet experiences of NZ users.

Perhaps more concerning and critical is the implications on security. If peering is simply left to posturing and shows of strength among the players (as seemingly encouraged on pages 16-17, the example of RadioNZ's streamed content), the traffic will as in this instance be routed overseas. Security of communications, and in particular Government and other commercial data, would be placed in a higher range of risk if forced to such overseas routing by gamesmanship among the Internet providers.

A move away from peering can, in general, terms, have the potential to in effect develop a two tier Internet in New Zealand. Diversity of choice for consumers through a robust ISP and content provider market should be encouraged, rather than simply having to face the 'market realities' that dominate when Tier 1 ISPs increasingly are able to control and manipulate the market.

The introduction to the Cook Report in late 2002 examined the experiences of peering in North America at that time
(http://www.cookreport.com/11.08-09.shtml).

Several of the paragraphs have relevance in New Zealand at this time:

"They [small ISPs] are weaving their positions at the edge of the network into a mesh of cooperation that is likely to provide resiliency for the internet as a whole as shockwaves from the collapse at the center propagate."

...

"They [larger Tier 2 players] are, in effect, climbing a peering "ladder" where as their bandwidth increases and they peer with each other and can get peering with larger players, they are likely to depeer with smaller players whom they feel they now no longer need and believe they can sell transit to. Thus although the plates of peering are shifting, the fundamental premie is likely to remain one where players peer only if their aggregate traffic is approximately equal. However peering is motivated as much by politics as economics. Therefore, it would be a mistake to think that these premises apply with equal force the world over."

InternetNZ believes that the paper does not adequately examine the wider range of motivations for and outcomes of peering and de-peering. While it addresses the economic and market positioning, it fails to look at social, user experience, societal good, encouragement of innovative content development and other factors that can and should be part of the peering equation. To leave the entire process open to the unfair manipulation of some of the participants will only ultimately result in a loss of participants in the market, and a loss to the end users in New Zealand, and perhaps the final outcome being the capture of the entire Internet user market by only a handful of service providers.

Document Actions