WP1 – Measurements: Mapping Manipulation in the Internet
The purpose of this work package is to perform large-scale measurements on the dynamics of the packet mangling by middleboxes in the current Internet. The measurements are performed in two phases: 1) an initial measurement phase using currently available tools and techniques on existing infrastructure (e.g. Onelab8 and Archipelago9) that will start early in the development of the selected MONROE testbed, but only allow for detection of middleboxes; 2) large-scale measurements using enhanced measurement and data collection techniques developed by the MAMI project on the FIRE+ testbed provided by the MONROE project and potentially other additional infrastructures (e.g., RIPE Atlas10), if suitable. The collected measurement data will be made accessible to the research community.
WP2 – Experimentation: Middlebox Modeling and Testing
In this work package, the measurement results from the studies performed in WP1 will be used to classify middlebox behavior observed in the Internet and to create models of how middleboxes mangle packets. These models will subsequently be used for experimentation with the mechanisms developed in WP3. Experiments will be performed in a large-scale testbed to evaluate the deployability of the Middlebox Cooperation Protocol (MCP), as designed in WP3, as well as any protocol extension that will be implemented in a flexible transport stack for middlebox detection and path selection. In addition the middlebox model will be used for NFV-based experimentation to evaluate the applicability of the MAMI approach to the selected use cases and thereby
its innovation potential for emerging services and network support.
WP3 – Architecture: Middlebox Cooperation
This work package focuses on the research, design, implementation, and initial testing of the Middlebox Cooperation Protocol (MCP). The MCP allows cooperating middleboxes to perform essential traffic manipulations appropriate to their function, while at the same time making accidental manipulation difficult. Together with mechanisms to detect middleboxes and perform a mapping between a suitable transport protocol and the path selection, the MAMI approach aims to restore protocol extensibility and thereby innovation by deployment of new protocols and protocol extensions in the Internet. To enable successful Internet-scale deployment of this flexible transport stack architecture this work package will in addition develop new encryption techniques to be integrated in the MAMI approach.
WP4 – Standardisation, Dissemination, and Exploitation
Two of the main objectives of this project, the design of a Middlebox Cooperation Protocol as well as its influence on NFV design and implementation, require standards as a basis for large scale deployment. Therefore standardization is a main focus of rthe MAMI project. Members of the consortium as well as the External Advisory Board are already working on pre-standardisation efforts in this space within the IETF (e.g. the SPUD BoF at the IETF 92 meeting in March 2015) and the IAB’s IP Stack Evolution Program (including the program lead and IAB membership). In what relates to NFV, members of the consortium are highly active in the ETSI NFV ISG (chairing its Technical Steering Committee, and the WG on Testing, Experimentation and Open Source), and the IRTF, co-chairing the recently created NFVRG.
WP5 – Project Management and Scientific Coordination
This work package especially fosters the scientific exchange between the partners and supervises the compliance to milestones as well as project objective.