|
NGI Implementation Plan
Section 2.2.1: Specifics/DARPA |
 
|
|
|
|
|||
 |
|
||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
2. Goal 1
|
Goal 1: Experimental Research for Advanced Network Technologies (continued) |
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2 Agency Specifics
|
Agency Specifics |
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1 DARPA
|
Defense Advanced Research Projects Agency |
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1.1 Introduction
|
Introduction DARPA's goals in network research mesh well with those of the initiative in key areas. Together with the terabit-per-second (Tbps) network speeds enabled by Goal 2.2 technologies, the network management and end-to-end QoS technologies developed under Goal 1 will be the pathway to a robust, scalable, shared infrastructure supporting lead DoD users, other government agencies, and the research community, as well as large numbers of commercial users. This will be accomplished through a partnership with industry. Specifically, the DARPA NGI program goals are:
|
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1.2 Network Growth Engineering
|
Network Growth Engineering The goals of this task are to (1) create and deploy tools and algorithms for planning and operations that guarantee predictable end-to-end performance at scales and complexities that are 100 times those of the current Internet; (2) facilitate management of large scale internetworks operating at terabit speeds supporting a range of traffic classes on a shared infrastructure; and (3) create an infrastructure partnership through which lead users share facilities with the general public to accelerate the development and penetration of novel network applications. This task will develop and integrate technologies for network planning and simulation; network monitoring, analysis, and control; innovative data delivery; and shared infrastructure management for lead users. The highly automated services envisioned in this task lead to the goal of building strong security mechanisms into the components. Planning and Simulation Planning large network interconnections is at present mainly a manual process that is not tied to any runtime tools or distributed efforts. Under this subtask, a network planning description language will be developed as a community standard. The target goals are to show that:
Network engineering and management requires tools for gathering and analyzing data and for issuing control commands based on the results. Current practice uses planned analysis based on protocol headers and aggregated statistics. This subtask will take on the challenge of presenting runtime analysis based on distributed communication patterns and communication speeds so high that they would overwhelm current tools. This will be accomplished by developing the following capabilities:
The goal of this subtask is to ensure that the network engineering tools and the high performance transmission and switching technologies work smoothly together for accomplishing end-to-end management of leading edge user requirements. The requirements will be developed with the users, and the tools will be integrated into the NGI testbed, developing 10 distributed management stations with monitoring and command interfaces to all connected equipment; half of the management stations will be in the Goal 2.1 testbed and half will be in the Goal 2.2 testbed. The management stations will demonstrate the scalability nature of the design and its capability to adjust to varying levels of granularity of information, organizational responsibility, and active versus passive monitoring and control. Analysis and control functions will work smoothly among cooperating organizations, and the ability to implement wide area, distributed interoperation will be demonstrated as an ongoing NGI effort. Data Delivery The management software will work in a tightly bound interlock with new strategies for controlling data delivery in networks. Research efforts under this subtask will develop network interior nodes that combine methods previously seen as disjoint or mutually exclusive: routing and switching, best effort and priority traffic, dynamic routing and virtual circuits, greedy admission versus guaranteed delivery, and flat rate versus variable costing. In addition, schemes that use generalized or alternative addressing methods will be explored. Tools that permit network engineers to adjust the strategy trade-offs to best meet their requirements will be prototyped and tested in the high speed arena. This subtask will develop the technology to allow lead users to share the same infrastructure as conventional users. The DoD, other government agencies, and the research community at large typically have lead user requirements for telecommunication facilities that are beyond those of the typical users by orders of magnitude. In some cases there may be sufficient aggregate capacity in the existing infrastructure to support their requirements; however, it is formatted or managed in ways that preclude coexistence among the lead and conventional users. Traditionally, wide area telecommunication service providers have addressed this problem by installing leased lines, a solution that has been extraordinarily expensive for the lead users. This subtask will investigate architectural concepts, management strategies, and operational arrangements that will facilitate the sharing of a common, wide area infrastructure. |
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1.3 End-to-End Quality of Service
|
End-to-End Quality of Service The goals of this task are to facilitate the delivery of end-to-end ensured QoS to applications and to ensure that these technologies can be tailored for use by lead users who have demanding requirements. The strategy is to allow users to negotiate application specific trade-offs among such parameters as bandwidth, latency, precision, and reliability in order to obtain predictable performance at a known quality level. Exploiting emerging network level mechanisms is difficult, however, for they are semantically far removed from the applications they are intended to support and are accessible only through layers of software. End-to-end QoS assurance requires an approach that spans these operating system and middleware layers in order to effectively deliver network level QoS guarantees. This task will develop and demonstrate a comprehensive QoS management architecture; drill down technologies to facilitate propagation of QoS constraints across software layers; and next generation network technologies to support QoS. Baseline QoS Architecture The baseline QoS management architecture will provide the framework of models, languages, and protocols to permit distributed applications to specify multidimensional QoS requirements, to negotiate acceptable trade-offs and confidence levels, and to receive feedback on delivered QoS enabling adaptation. APIs supporting the propagation of QoS constraints and feedback through software layers will be developed. The baseline architecture will support a broad range of defense, government, and commercial applications by focusing on the fundamental QoS dimensions of performance, availability, precision, and soft real time. This will subsequently be extended under DARPA funding to cover mission critical properties. Specific areas to be investigated under this subtask include:
Current technologies support composition of functionality across system layers, but not the composition of their QoS properties. This task will develop technologies that can be used to drill down and expose interfaces to QoS and network management capabilities that are presently hidden within the individual layers. Emphasis will be on techniques appropriate for operating systems, communications libraries, and middleware services, including distributed objects. Specific topics to be addressed include:
|
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1.4 Security
|
Security The NGI security goals are to provide the basis for implementing and enforcing appropriate security policies among organizations, users, and infrastructure components under shared control. Interoperable authentication methods are a prerequisite. NGI will meet the continuing challenge of building new services that use the network security architecture, but it will also develop ways for organizations or individuals to interoperate in the face of a rich and dynamic set of policies, for example, those that might exist among different Federal agencies. The assurance that security mechanisms are available, correct, and used will also be addressed. DARPA's primary security activities under NGI will be in securing the network management functions, exploring the integration of security into the QoS architecture, and ensuring the secure activation of drill down mechanisms. |
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||
|
2.2.1.5 Milestones |
Milestones Network Engineering
Quality of Service
|
||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||