NeTS-FIND: Collaborative Research: Towards Complexity Oblivious Network Management

NetS-FIND：协作研究：迈向复杂性不经意的网络管理

• 批准号: 0626978
• 负责人: Paul Francis
• 金额: $ 30万
• 依托单位: Cornell University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-09-01 至 2009-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0626978&HistoricalAwards=false
• 关键词: NeTS; FIND; Collaborative; Research; Towards

Proposal Number: 0626978PI: Paul T FrancisInstitution: Cornell UniversityProposal Number: 0627086PI: Jay LepreauInstitution: University of UtahTitle: NeTS-FIND: Collaborative Research: Towards Complexity-Oblivious Network ManagementAbstractNetwork management is costly and error prone. At best, this imposes a tremendous cost on industry. Worse, it compromises critical infrastructure that increasingly relies on the Internet, such as power and emergency services. The difficulty of network management derives from the fact that the Internet was invented without a network management architecture. Rather, there are a plethora of after-the-fact management tools and objects.This project is a ground-up redesign of network management based on three core concepts. First is a low-level, self-configuring communication channel that allows software network managers to discover devices and other managers. This channel allows the management system to function even when higher level protocols do not. Second is an abstract interface that allows network modules to represent themselves in simple ways. This abstraction allows management systems to understand the network without understanding the details of protocol operation. Third is a declarative configuration language that allows human managers to specify network operation in goal-oriented terms. This language allows managers to focus on what the network is supposed to accomplish, instead of how.This research will reduce the manpower and expertise needed to manage networks, increase their reliability, and ultimately lead towards largely self-managing networks.This research will be tested and deployed in Emulab, a large network testbed used by thousands of researchers. Methods hardened in Emulab can serve as the basis for management of the GENI infrastructure, where the ideas can be further tested and refined. GENI in turn will serve as a proof-of-concept for industry adoption.

建议编号：0626978PI：Paul T Francis Institution：Cornell University建议编号：0627086PI：Jay LepreauInstitution：Utah University of Utah标题：Nets-Find：Collaborative Research：走向复杂性忽略的网络管理摘要网络管理成本高昂且容易出错。在最好的情况下，这给工业带来了巨大的成本。更糟糕的是，它损害了日益依赖互联网的关键基础设施，如电力和紧急服务。网络管理的困难源于互联网是在没有网络管理架构的情况下发明的。相反，有太多的事后管理工具和对象。这个项目是基于三个核心概念的网络管理的彻底重新设计。首先是一个低级别的、自我配置的通信通道，允许软件网络管理器发现设备和其他管理器。该通道允许管理系统即使在更高级别的协议不起作用时也能起作用。第二个是抽象接口，它允许网络模块以简单的方式表示自己。这种抽象允许管理系统在不了解协议操作细节的情况下了解网络。第三种是声明性配置语言，允许管理人员以面向目标的术语指定网络操作。这种语言使管理人员能够专注于网络应该实现什么，而不是如何实现。这项研究将减少管理网络所需的人力和专业知识，提高网络的可靠性，并最终实现网络的高度自我管理。这项研究将在Emulab进行测试和部署，这是一个由数千名研究人员使用的大型网络试验台。在Emulab中强化的方法可以作为GENI基础设施管理的基础，在那里可以进一步测试和改进想法。GENI反过来将作为行业采用的概念验证。