|
NGI Implementation Plan
Section 3.1: High Performance Connectivity |
 
|
|
|
|
|||
 |
|
||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
3. Goal 2
|
Goal 2: NGI Testbed |
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1 Goal 2.1
|
Goal 2.1: High Performance Connectivity |
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.1 Introduction
|
Introduction NGI Goal 2.1 will perform R&D on advanced interconnection technology to develop and deploy a wide area demonstration network fabric that will support network research and advanced network capable applications as defined in NGI Goals 1 and 3. This fabric will function as a distributed laboratory delivering a minimum of 100 times the current Internet performance on an end-to-end basis (typically greater than 100+ Mbps end-to-end) to at least 100 interconnected NGI universities, national laboratories, and Federal research sites that demonstrate research and applications that require such an infrastructure. This demonstration network fabric will be large enough to provide a full system, proof-of-concept testbed for hardware, software, protocols, and network management that is required in the commercial NGI. It should be noted that the emphasis of Goal 2 is on terrestrial networking technologies and connectivity. Wireless technology is becoming an important element of tomorrow's Internet architecture. However, because of the intense competition in the wireless industry and insufficient government resources, NGI will highly leverage industry developments and any ongoing Federal research to provide a hybrid networking demonstration platform. The following sections describe the NGI Goal 2.1 implementation plan for the agencies included in the Congressional FY 1998 NGI appropriations. DoE is not a formal participant in the NGI in FY 1998. The Administration plans to propose adding DoE as a formal participant beginning in FY 1999. DoE's participation in the NGI beginning in FY 1999 is described in Appendix E. |
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2 Strategy and Subgoals
|
Strategy and Subgoals To implement this goal, strategy and metrics will be defined for each of the following subgoals:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.1 Infrastructure Subgoal
|
Infrastructure Subgoal The large scale development of advanced applications calls for a network that is relatively stable in order to provide a consistent environment to which developers can build, and one that is also leading edge in order to incorporate new technologies as soon as they are available. Development and testing of the new network technologies requires a more flexible network, one that can be modified and tested frequently. These objectives will be met in NGI Goal 2.1 by expanding and interconnecting existing Federal research networks into a large "leading edge but stable" network of networks for about 100 sites to test specific technologies and projects. This is envisioned as an advanced network infrastructure that supports the research identified in Goal 1, but that will also strive to provide a stable network to the applications. However, there will be occasions when the network may suffer partial degradation of services to support these sometimes divergent goals. A balance must be achieved between a stable "production" environment and one that allows for research and new technology insertion so as to enable both. For the most part this can be accomplished with existing technology. It is envisioned that: (1) a relatively stable but leading edge infrastructure will be developed that will have some downtime as new technologies are developed and introduced; (2) sites will have backup commodity service; (3) a service definition will exist between the end user and providers -- this will include such items as latency, demand, and scheduled service. The strategy to form the leading edge but stable network is to connect the university community (including, for example, leading research universities and Internet2 campuses) with the vBNS national backbone, connect the Federal laboratories and other research sites with agency research networks, and interconnect them at high performance exchanges. These networks will seek to provide a stable infrastructure, but may potentially be less stable as they aggressively implement the research, development, and integration of new technologies and applications being developed in Goals 1 and 3. This strategy allows for sharing the costly underlying carrier infrastructure while providing a network that the new 100x technologies can be deployed in a limited but real world environment for testing before being introduced into the full 100x testbed. The performance metrics that will be used to determine if the infrastructure subgoal has been achieved are:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.2 Common Bearer Services Subgoal
|
Common Bearer Services Subgoal In order to create a relatively stable infrastructure that can be quickly deployed, IPv4 with "best effort" services will be the initial and minimum common bearer service for all agency network infrastructures. IPv6 and other advanced technologies for QoS, etc., will be deployed on the networks as soon as they demonstrate a "leading edge, but stable" performance. Direct connections in ATM, IP over SONET, and other services will be introduced on the basis of feasibility, support for Goal 1 research activities, and application demand. The use of virtual backbones over the existing agency networks to perform the required R&D will assist in developing these technologies and services, provide a relatively leading edge but stable infrastructure for applications development, and assist in solving problems of interagency and intercarrier connectivity at the ATM level. The metrics to determine success of this subgoal will be to gather statistics on a number of connected sites, and to keep track of trends of the bearer services in use and the number of sites that transition to advanced services, that is, from IPv4 to IPv6 and ATM. |
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.3 Application Feedback Subgoal
|
Application Feedback Subgoal One of the basic failures of previous high end testbeds was that although the infrastructure may have been a success, the initiative did not attain its potential because there were virtually no applications that could effectively use the infrastructure. This subgoal will help to ensure that this does not occur in the NGI initiative. Not only will NGI site selection be based on the availability of NGI class applications, but it will also include application developers working with network researchers to ensure that the resulting infrastructure satisfies the application requirements and enables those applications not yet conceived. Applications developers will be intimately involved in the R&D and deployment of the NGI. The strategy that will be used to facilitate applications development and migration onto the 100x testbed is to provide feedback from application developers to network researchers, operators, and implementors by means of regularly scheduled coordination meetings and conferences among these groups. It is envisioned that coordination meetings will occur at least four times per year over the life of the NGI initiative. The coordination conference responsibility will rotate among the agencies involved in the Goal 2.1 testbeds. All coordination conferences will be broadcast on the multicast backbone (MBone) to obtain maximum information dissemination. In addition, common links to up-to-date Web pages will be maintained by the networks and agencies involved. At a minimum, these pages shall contain information on the various applications under development on the 100x testbed, current status of the applications, problems encountered and solutions, and action items from the various meetings and conferences. In addition, conferences that address specific issues or topics will be held on an as needed basis. The responsibility of holding the conferences will fall to the agency or group concerned with the particular issue. In addition, to continue successful transition from the Federal testbed sector to the private sector and to ensure that applications developers have continued infrastructure support, a transition coordination group will exist for some time after the conclusion of the NGI. The metrics that will be used to determine if this subgoal has been met are:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.4 Interconnection Subgoal
|
Interconnection Subgoal In order for the NGI initiative to lay the foundation for the future Internet, it is imperative that the Federal networks provide end-to-end QoS, interconnecting with advanced services at the IP and lower layers as appropriate. It is also recognized that different levels of QoS will exist as dictated by the needs of the applications, for example, e-mail does not require the same QoS as does a real time collaborative environment used to control an experiment at a remote site. Implementation of NGI Goal 2.1 will create a nationwide seamless NGI fabric that provides appropriate levels of QoS, as well as necessary internetwork management security. Each agency will leverage its existing virtual domains (that is, administrative domains with different centers of control). The agencies have already involved the major carriers. One goal of the NGI is for these carriers to cooperate seamlessly on interconnection of the Federal agency networks. To facilitate this, the agencies will form an open interagency carrier and switch vendor working group to facilitate interagency and intercarrier cooperation at the implementor level. This group will also provide input to the agencies to help define the phases for seamless interconnection across domains (that is, IP, Permanent Virtual Circuit-PVCs, etc.) as well as transition from carrier-based to interconnection-based services. The metrics that will be used to determine success are:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.5 Site Selection Subgoal
|
Site Selection Subgoal Since NGI research crosses sites and agencies, a coordinated mechanism is needed to ensure that the sites have valid applications and the required infrastructure to promote the NGI Goal 2.1 success. To achieve this, each agency will identify its NGI sites and will work in a coordinated manner with the other agencies and the application group in the site selection process. The technical engineering and design issues will be addressed by the existing LSN's Joint Engineering Team (JET). The sites with valid requirements, that is, Goal 1 and NGI applications research, will present their proposals to the respective agencies that will determine the merits, based on reviews, and award the required funding or provide the required services or access. The agencies will be responsible for ensuring that a phasing plan exists in the proposal, where appropriate, for the site to migrate to production services with non-NGI funding in the near future. They must also ensure that connection to GigaPOPs is present where required; that cost, local issues, carrier, and site infrastructure plans are in concert with the NGI Goal 2.1; and that support for the NGI applications exists. The metrics that will be used to determine success of this subgoal are:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.6 Network Management Subgoal
|
Network Management Subgoal In order to make NGI Goal 2.1 a success, the various testbed operators must be able to ensure that the testbed is manageable and performing to the required levels. To achieve this, network management must be addressed early in the NGI initiative. It is envisioned that the agencies will have distributed, yet coordinated, network management efforts that build on existing agency efforts. The key features of this network management effort are:
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
|
3.1.2.7 Information Distribution and Training Subgoal |
Information Distribution and Training Subgoal One of the key items necessary to ensure that an initiative succeeds is to provide a mechanism for information dissemination and training. The effort under this subgoal is to ensure that information distribution occurs on a regular basis and that mechanisms are put in place to provide the required level of training to ensure that the NGI can succeed. These efforts will consist of, but are not limited to:
|
||||||||||||||||||||||
|
|
|
||||||||||||||||||||||
|
|
|||||||||||||||||||||||
