Throughout the history of the Internet, the evolution of networking protocols and related technologies has been tightly coupled to the research community. SWITCH is keeping this tradition alive by supporting projects in the field of network research.
A prominent example of the influence of the research community on the design of fundamental protocols is the following: in 1988, Van Jacobson, then at the Lawrence Berkeley National Laboratory, identified and fixed a fatal flaw in the early TCP congestion control algorithm. It had previously caused multiple crashes in most parts of the Internet that existed at the time . By applying scientific reasoning and elements of control theory, he came up with a small set of modifications to the TCP/IP code, that reduced the throughput of data transmissions in early stages of congestion to avoid a total and permanent breakdown of the entire network. His algorithm is still part of the inner workings of the current Internet.
Research in the realm of communication networks has long since become a discipline of its own, with countless sub-disciplines and a vast output of scientific papers. There is a constant urge to improve protocols, fix perceived flaws or even completely redesign the core of the Internet itself. One of the challenges in these endeavors is to bridge the gap between pure research and operational reality. After all, the current network has to be built, configured and operated mainly on principles of cost efficiency, reliability and manageability in complex business environments.
Established in the early, research-centric Internet, SWITCH has a long tradition of collaborating with national and international research projects. In fact, we believe that the willingness to share information about our network as well as to actively support projects by, for example, providing insights into our daily operations is an important part of our mission.
There is a certain risk for researchers to fall victim to the ivory tower syndrome: For example, they may work on problems that are technically challenging but have little relevance in practice. Or making assumptions about network characteristics like traffic patterns, network topology or routing policies that do not reflect reality. The value that SWITCH can provide in this context is empirical data from a real network and experience in managing an autonomous system in the global routing infrastructure.
In turn, SWITCH can benefit from the researchers, for example by gaining new insights into its own network through novel methods of data analysis.
Two recent projects may serve as examples for such fruitful interactions. The Networked Systems Group at ETH Zurich, lead by Laurent Vanbever, is, among many other topics, working on a system for inferring network behaviour from forwarding state and router configurations from the running system. It allows the operator to pose questions about his network in natural language. Such a tool could significantly improve the ability of an operator to understand, improve and manage his network.
The second project comes from the Institute of Information Security, also at ETH Zurich. Adrian Perrig and his team are working on a new inter-domain routing protocol that provides control over network paths by end-systems to protect users from certain malicious or accidental routing failures. Additional services include the simultaneous use of diverse paths, rapid failover and dynamic traffic optimization. SWITCH is part of the test-bed where this new protocol is being evaluated in a semi-production environment.
By working together, learning from the past and welcoming new ideas and concepts, we can play our part in helping to shape the future of the Internet.
 SIGCOMM '88 Symposium proceedings on Communications architectures and protocols, pages 314-329