wiki:OldGPGCommunityParticipation

*GENI PLANNING GROUP DOCUMENTS ARE NO LONGER CURRENT. See GpoDoc and the GENI project pages for up-to-date documents and specifications.*

Broad Participation

While we describe GENI primarily as an NSF initiative, our expectation is that GENI will experience broad participation, and in fact, its design explicitly fosters such participation. Industrial

The networking research community has a long history of support from the computing and communications industry. For example, the experimental facilities that exist today have support from corporate partners as diverse as Intel, Hewlett Packard, Google, AT&T, Cisco, Microsoft, Nortel Networks, Lucent, and DoCoMo. Likewise GENI will also seek active participation from industry. This participation will take two principal forms.

The first form is straightforward, based on standard contracting relationships. We envision that many elements of the GENI facility will be constructed using contracts or sub-contracts with industrial partners. Depending on the work to be performed, such contracts might be with networking equipment vendors, electronic device designers and low-volume manufacturers, bandwidth providers, and network operators (ISPs), among others. In all cases, however, this class of relationship will be marked by a clear specification of the work to be performed.

The second category of participation is both more open-ended and more interesting. We envision the likelihood that industry partners will choose to contribute to the GENI facility for a variety of reasons not related to immediate contractual reward. These reasons might include

  • More rapid deployment of technologies developed within the industry. Should GENI be successful as an early-deployment catalyst, there is no reason it should not serve this role for new industrial technologies as well as new research results.
  • A more direct path for moving research results to commercial practice. Here again the benefit derives from GENI's role as an early deployment enabler. In this scenario, however, the deployment being enabled is that of new research results on top of industrial platforms. In essence, by designing and implementing commercial equipment that fits within the GENI model, the likelihood of new research results moving rapidly from the GENI infrastructure itself to the broader commercial setting is increased.
  • Relevant industry partners will receive some incentive to contribute to GENI for a variety of "standard" reasons -- public visibility and public relations, tax considerations, and the like.

Many of the steps to lower barriers to continual renewal (the requirements for GENI) are also those needed to lower barriers to industrial partnership. The appropriate use of modularity, well-defined class-based interfaces, and cleanly separated global and device-specific functions combines to greatly simplify the task of integrating a vendor's products into the GENI architecture and facility. To further lower barriers to integration, we identify two further, related, requirements. These are stability of the defined interfaces, and the existence of feedback mechanisms for clarifying, enhancing, and evolving these interfaces.

Taken together, these requirements make the integration of vendor-developed components into the GENI facility as simple as possible. A well-defined, well-documented, stable architecture and interfaces that impose minimum requirements on the vendor provide the basic framework for integration. A process for transmitting understanding of these interfaces from the GENI designers to the vendor community and for providing and incorporating feedback from the vendor community to the GENI project closes the loop, lowering technical barriers to industry participation to the greatest possible extent.

International

GENI is designed to provide global reach, making international participation essential. While some international reach can be provided by GENI proper (e.g., by leasing co-location space around the world), our expectation is that other countries will build GENI-like facilities and we expect GENI's design to support a federation of such facilities. This is already happening through international participation in PlanetLab, which includes as many sites outside the U.S. as within. Many of these sites are connected to (and sponsored by) the hosting country's national research network. This is true in Canada and throughout Europe, as well as in Japan, Brazil, India, and China.

GENI will be explicitly designed to support federation, which gives each participating organization autonomous control of its own resources. This will allow GENI to both accommodate existing facilities, as well as expand over time as additional countries decide to participate.

Government Agencies

We expect significant participation from additional research communities, application domains, and the government agencies that support them. We envision this participation happening in two ways.

First, the basic ideas behind GENI have been presented to the National Coordination Office for Networking and Information Technology Research and Development (NCO/NITRD), an organization that coordinates networking R&D activities across various government agencies. The NCO/NITRD reports to the President's Office of Science and Technology Policy (OSTP). We will continue to develop GENI in consultation with this Office.

Second, we envision individual research groups and funding agencies wanting to connect their experimental facilities to GENI, thereby gaining access to new services and functionalities that are being deployed. For example, it is easy to imagine a scientific community using GENI's sensor network building block as a reference implementation for a purpose-built sensor network. It should be easy to connect this sensor network into GENI. As another example, one can imagine a community with high-bandwidth needs gaining access to one or more lambdas wanting to integrate management of this capacity into GENI for the sake of taking advantage of GENI-provided services. GENI's support for federation, as well as its ability to absorb new networking technologies, will enable these sorts of activities.

Last modified 15 years ago Last modified on 04/30/09 12:27:04