Abstract

Next generation network architectures will benefit from the many years of practical experience that have been gained in designing, using and operating network protocols. Over time, the networking community has gradually improved its understanding of networked systems in terms of architecture, design, engineering and testing. However, as protocols and networked systems become more complex, it is our contention that it will be necessary for programming techniques to evolve similarly so that they better support the design, implementation and testing of both the functional and the non-functional requirements for the network protocols that will be used. We therefore envisage new levels of programming language support that permit: (a) the design and implementation of new protocols with provably correct construction; (b) inline testing; and (c) the expression of protocol behaviour within the design. Based on our ongoing work with both network protocols and programming language design, we believe that exploiting the capabilities of recent work in Domain Specific Languages (DSLs) will allow us to meet such requirements, allowing straightforward and “correctby-construction” design and implementation of next generation network protocols. 1.

Citations