NOVI is new research project that aims at developing a federation of “Experimental Facilities”. This is indeed needed looking at the very many “test beds, pilot sites, experimental infrastructures and facilities” that were built in the past with public funds and co-innovation research all over Europe. Without knowing numbers, there must be hundreds if not thousands of such sites that allow to experiment with network architectures and more recently also compute and storage systems.
What remains to be seen is whether a cohesive/coherent federation will lead to an upsurge of usage, especially by industrial research organisation. From this perspective, the issue is less scale and simplicity but rather legal aspects related to running code on nodes/hosts that are operated by someone in the “experimental cloud”.
Some time ago I had a closer look into PlanetLab, perhaps the pioneer in this domain, and the concepts behind truly appeal. Technically, the conceptual proximity to Infrastructure-as-a-Service offerings, like the one by Amazon or Akamai, is intriguing. And usage statistics indicate significant interesst by the community. The biggest disadvantage in my view was/is the lack of support for reproducability of experiments as resources are not explicitely granted and isolated. Essentially, one can conduct an experiment at large-scale and realistic conditions but each and anyone remains unique and hence (striclty) uncomparable.
This is supposted to be different for GENI, which is a pretty large-scale infrastructure for a “Future Internet” (communication architecture). Some claim it will turn sooner or later, once the Future Internet Architecture is identified, into the Future Internet, just like the ARPA/DARPA Net did orignially.
So something that is worth to keep an eye on.
But along with Cloud Computing (Infrastructure-as-a-Service) offerings the IT domain is also looking more into this domain and OpenCirrus is a perfect example for this. It remains a somewhat semi-pubilc resource but allows to experiment with IaaS down to the virtualization layers (and slightly beyond).
It”s hard to keep track of this domain, indeed. In Europe there is a whole research theme / community after it – Future Internet Research & Experimentation – but the needed for experimentation seems to be there. Actually I know too little of actual usage stats and it was something that called experimental facilities in question over the past.
But now I got invited to the advisory board of the NOVI project and I am truly looking forward to gaining deeper insight in this domain.