The SESERV Coordination Action and the SmoothIT project organized a joint panel on “Economic and user perspective of Inter-ISP traffic optimization”. The panel had a slot of one hour. It was moderated by Martin Waldburger (SESERV, University of Zurich). The panel saw the following panelists, who are experts in inter-ISP optimizations by their respective research and project backgrounds:

Nicola Giulli (ETICS, Head of R&D activities at Nextworks): In his statement slides entitled “Discussion seeds from ETICS”, Nicola outlined business-, strategic, and technical obstacles to be overcome for inter-ISP optimizations. He also emphasized on the erosion of transit prices that may render inter-ISP optimization approaches obsolete due to a management-wise much simpler to achieve over-provisioning becoming feasible.

Frank Lehrieder (SmoothIT, University of Würzburg): Frank focused in his statement slides on the BitTorrent case, which reflects a context in which he concludes considerable reduction of inter-ISP traffic being possible when making use of the right ETM (economic traffic management mechanism), e.g., when taking locality information into consideration in the overlay. While ISPs have a clear interest in reduced costs by these optimizations, there are no such clearly visible benefits for P2P users. Frank, thus, concluded that ISPs have to provide incentives for users (e.g., financial rewards, improved QoE, improved Internet access). Without the respective user incentives, Frank said, inter-ISP optimizations will not work.

Filip de Turck (OCEAN, Ghent University): Filip emphasized on the implications of a Cisco forecast that predicts 91% of inter-ISP traffic in 2014 will be video data. Accordingly, he stated that the respective QoE metrics need to be taken into account in inter-ISP SLAs. He also outlined that not only multi-hop QoE agreements will be needed, but also the right Inter-ISP resilience mechanisms have to be in place.

Burkhard Stiller (SESERV, University of Zurich): Burkhard outlined, based on the example of a traffic map, where inter-ISP optimizations potentially can take place and how the load situation typically looks like. He then listed the relevant set of stakeholders involved and concluded by the analysis of their respective interests that there are multiple diverging interests among all players. This situation leads to tussles between ISPs, users, and application providers. Out of the tussle space outlined, he concluded that users do not care much about inter-ISP issues. Inter-ISP optimizations seem to be essentially in the lone interest (in terms of incentives provided) of ISPs only.

For more details on all panelists, their research background interests, as well as the overall panel schedule, see the panel overview presentation. The moderated discussion following the statements round led to the main three insights as follows:

• Users are very difficult to involve in inter-ISP optimizations. They have the power to influence an ISP’s cost structure by the choices they make, such as the use of overlay nodes in- or outside an ISP’s network.

• The multi-hop inter-ISP optimization case might not be equally important to all stakeholders and for all application scenarios.

• There might be a trade-off between the economic approaches to inter-ISP optimizations (resulting in management effort for the purpose of total cost reduction) and the more simplistic over-provisioning case (which may become more attractive due to eroding transit prices).