Despite or because of the Internet?s great success, it has shown itself to be very resistant again attempts to add new functionality to the network core. This dilemma lead to the recent trend in networking to implement services using overlay networking; overlay networking provides an opportunity for end systems to collaborate with others to achieve enhanced functions without having to modify routers. However, since overlay networks operate at the application layer, they cannot effectively use the resources that are available to network services. To address the limitations of services in the current Internet, this work presents a clean slate Internet architecture, called NetServ, based on the concepts of service virtualization. NetServ strives to break up the functions provided by Internet services and to make these functionalities available as modular building blocks for network services. A building block effectively encapsulates a network resource or function realized by a network node, such as link monitoring data or routing tables, and provides tunable parameters for easy configurability. These building blocks form a foundation for the implementation of full-fledged network services, which can then be used by applications. Furthermore, the new framework handles aspects of service discovery, distribution and management, thereby making new services more readily deployable. This project addresses five major research challenges in the new service architecture: i) the definition of requirements for a service-virtualized Internet architecture, ii) the design of an architectural framework for modular, virtualized services, iii) the identification of an initial set of key building blocks, which together can provide a foundation for common network services, iv) the development of mechanisms and protocols for service discovery and service distribution, and v) the design and implementation of a content distribution service based on the NetServ architecture. The feasibility of the NetServ approach will be demonstrated by building the prototype of a content distribution service. This service will interwork with two other services that are crucial for a variety of applications: a generalized naming service and a network monitoring service. The naming service is capable of naming a variety of network entities (like content, users, services, and devices), while obeying policy constraints. The network monitoring service provides makes network performance data available to applications. Overall, the objective of this work is to develop an architecture that provides an efficient and extensible architecture for core network services, to implement a prototype of the new architecture, design and implement a prototype of a content distribution service to demonstrate the feasibility of NetServ, and to evaluate the new architecture in a simulation environment as well as on a GENI-like test bed. Intellectual Merit: NetServ fully decouples service execution from packet forwarding on a network node. A service only assumes that a node provides packet transport and possibly offers building blocks that can be used to access resources. The new architecture for next-generation Internet services, therefore, provides an efficient and flexible foundation for network services, makes the future Internet highly evolvable, and decouples services from a specific technology. Broader Impact: This work will make the Internet service-centric and enable services to directly utilize network resources as building blocks. NetServ can serve as an excellent container for integrating other ongoing FIND work at other institutions. This project will also contribute to both the graduate and undergraduate experimental and theoretical networking curriculum, and involve under-represented groups within its research efforts.
