Service introduction in an active network

In today's networks, the evolution of wide-area services is constrained by standardization and compatibility concerns. The result is that the introduction of a new service occurs much more slowly than the emergence of new applications and technologies that benefit from it. To ameliorate this problem, an active network exploits mobile code and programmable infrastructure to provide rapid and specialized service introduction. A viable active network has the potential to change the way network protocols are designed and used, stimulating innovation and hastening the arrival of new functionality. There are, however, a number of challenges that must be overcome in the design of an active network. Chief among them are how to express new services as network program, and how to execute these programs efficiently and securely. In this thesis, I present a novel network architecture, ANTS, that tackles these challenges, and describe its prototype implementation in the form of a Java-based toolkit. The main finding of this research is that ANTS is able to introduce new services readily and at a reasonable cost. Experiments with multicast and Web caching services provide evidence that new services can be constructed despite a restricted programming model and constraints such as a network that is only partially active. Measurements of the toolkit show that the forwarding mechanism requires little processing beyond that of IP, such that even the user-level Java toolkit is able to run at 10 Mbps Ethernet rates. Moreover, ANTS achieves this efficiency while raising few new security concerns compared to IP, the most significant of which is that the global resource consumption of a service must be certified as acceptable. I conclude that the active network approach shows great promise as a means of promoting network evolution. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)