Is There Life After the Internet?

Computing paradigms have historically evolved through massive disruptive changes: from mainframe-computing to desktop-computing, to Internet-computing. It is thus not unreasonable to ask whether Internet-computing may be similarly transformed, and if so, what might be the characteristics of post-Internet computing. This presentation considers these questions in terms of three concomitant technology changes: (a) Massive storage growth at clients and its respective download-&-play applications are stretching the Internet's client/server-unicast paradigm to its scalability limits. Future networks will need to scale to multi-TB personal storage system, well beyond the scope of current paradigms. What technologies and applications paradigms may emerge to support these new scales? (b) Tagged XML software is inverting the classical sequential-code/random-access-data computing into sequential-data/random-access-code; this enables flexible dynamic distribution of data and processing logic of network services beyond the scope of traditional centralized processing. For example, it is not improbable that future services will replace centralized delivery by Web sites with distributed personalized storage sites. (c) Traditional architectures have organized processing resources to scale I/O-memory speeds to exploit the much higher speeds provided by CPU technologies. Emerging wire-speeds and storage-speeds have been inverting the classical speed hierarchy, with I/O rates outstripping CPU speeds. There is a need for new HW/SW architectures to handle the scalability challenges presented by this inverted speed hierarchy.