FIRE Portfolio Analysis (spring and summer 2010) Executive Summary

The complete portfolio analysis has been carried out under the FIREWorks support action by Scott Kirkpatrick and Jerker Wilander (editors), Jacques Magen, and Dirk Trossen. This Executive Summary is released by the FIRESTATION support action.

With the FIRE portfolio the opportunity now exists to support new classes of users and experiments combining heterogeneous technologies, all representing key aspects of the future Internet. The research done already in FIRE and starting in the current Call compares well with the best work in the world, including the highly visible GENI portfolio in the US and national projects in the Far East. However, the FIRE projects to date have yet to set up a federated testbed for European researchers. This can only be achieved by overcoming the considerable obstacles of complexity and unfamiliarity that are faced when trying to explore the effects of new applications that bring future users the increasing power of the future Internet, from its edge to its core with computation and storage embedded everywhere. Individual projects, for the most part, have stressed creating improvements in one technology, or two technologies that they can connect directly; but federation has not yet been considered as an issue they would all have to deal with.

FIRE differs from GENI in that FIRE emphasizes the value as seen by an end-user, at the edge of the Internet, with its applications or services, while GENI is focusing more on basic infrastructure technologies.

The usage of individual FIRE facilities has not been as extensive as originally expected among FIRE call 2 projects. The facilities that have been used to a reasonable level and not only for FIRE internal usage are OneLab’s PlanetLab Europe and FEDERICA. WISEBED and Panlab II were used extensively by their own project partners, however, did not manage to attract users from outside their own communities as they were not made available outside their consortia. Uncertain sustainability will not ease further usage. In addition the requirement for stronger user support and better dissemination of availability is important.

Existing FIRE facilities could have been used by some of the research projects, but in most cases the choice was to either use other existing test beds or building a small test bed for internal experimental use. Further dissemination of information about the availability and details about the offerings of FIRE facilities should have prevented this.

Several FIRE supported projects show that there is a need for an experimental facility supporting emulation. This is emphasized by the use and planned use of the “IBBT iLab.t” (exemplified by OFELIA, CREW, ECODE and Euler).

A major step forward on the understanding of what can be performed on existing experimental facilities is expected from the FIRE day in Ghent in December 2010, when Call 2 projects plan to present their major use cases as showcases.

Some specific actions needed to improve the FIRE experimental facilities are:

FIRE should address the integration the top-down federation in FIRE projects (mainly through PII/Teagle) and the bottom up federation as represented by Sliced-Based Federation Architecture/SFA (in FIRE + GENI) which show approaches with different properties.
The issue of how the NREN and GEANT infrastructures can be used in FIRE should be discussed and concluded. This issue has been discussed since the beginning of the FIRE initiative. Experience in OneLab, FEDERICA, and OFELIA can serve as a starting point.
The sustainability of each FIRE experimental facility should be studied in the context of a sustainable FIRE federated facility. It is important that the FIRE Architecture Board, set up by the FIRESTATION support action, investigates during the coming months proper business models, using the model of contributing and sharing resources and more commercial models as a starting point. Considerations about how far networking costs must be paid for by the user should also be considered taking into account the experience gained in the FEDERICA project.
The FIRESTATION support action shall take the lead in identifying appropriate levels of user support and ensuring that the best practices are shared.Benchmarking and repeatability of experiments is important when conducting experiments of high quality and scientific impact. This will put requirements on sharing/interconnecting experimental data across experiments and supporting comparable measurement capabilities, not only resources. Standards and shared tools in this area should be organized once the shape of the experiments performed under the open calls is visible.
Last but not least, we believe that this overall coordination needs to be streamlined in the light of the upcoming FI PPP and ICT Calls 7 and 8. Less is more! One opportunity in this context is the Call for a FIRE Federation IP under Call 8.