Internet as a Dataflow Computer
This paper proposes a novel dataflow architecture called DFNET (DataFlow architecture on the interNET), which realizes a scalable dynamic dataflow computer on a packet switching network. A vast amount of research on dataflow computers has been extensively performed during, in particular, 1970s and 1980s as a promising approach for realizing very high-speed computers. In spite of the large amount of expectations, success of dataflow computer researches is quite limited. The objective of this paper is to present the concept of a scalable and extensible dataflow architecture on a packet switching network, which utilizes the abundant resources of a large-scale computer network. In this paper, we introduce the concept of DFNET (DataFlow architecture on the interNET). DFNET is composed of configuration and control methodologies of routers in a packet switching network. The key of DFNET is that a packet switching network is utilized not for end-to-end communication but for dataflow computing. Because of desirable properties of a packet switching network, DFNET has high scalability in terms of the dataflow program size and high robustness against failures.
KeywordsProgram Execution Virtual Link Upstream Node Downstream Node Output Interface
Unable to display preview. Download preview PDF.
- 1.Sharp, J.A. (ed.): Dataflow computing: theory and practice. Ablex Publishing Corporation (1992)Google Scholar
- 3.Dennis, J.B., Misunas, D.P.: A preliminary architecture for a basic dataflow processor. In: Proceedings of the 2nd Annual Symposium on Computer Architecture, pp. 126–132 (1975)Google Scholar
- 4.Baker, F.: Requirements for IP version 4 routers. Request for Comments (RFC) 1812 (June 1995)Google Scholar
- 6.Bhattacharjee, S., Calvert, K.L., Zegura, E.W.: An architecture for active networking. Tech. Rep., College of Computing, Georgia Institute of Technology, Atlanta, Georgia (1996)Google Scholar
- 7.Gostelow, A.K.P., Plouffle, W.: An asynchronous programming language and computing machine. Tech. Rep. 114a, Department of Information and Computer Science, University of California, Irvine, CA (1978)Google Scholar
- 8.Gurd, J.R., Watson, I.: A multilayered dataflow computer architecture. In: Proceedings of 7th International Conference on Parallel Processing (August 1977)Google Scholar
- 10.Suzuki, H., Ohsaki, H., Sawai, H.: Algorithmically Transitive Network: a new computing model that combines artificial chemistry and information-communication engineering. In: Proceedings of the 24th Annual Conference of Japanese Society for Artificial Intelligence (JSAI), pp. 2H1–OS4–5 (2010)Google Scholar
- 11.Suzuki, H., Ohsaki, H., Sawai, H.: An agent-based neural computational model with learning. In: Conference Abstract of the 3rd INCF Congress of Neuroinformatics (Neuroinformatics 2010) (2010), doi:10.3389/conf.fnins.2010.13.00021Google Scholar