Proxy-based Dissemination of Dynamic Web Data
University Of Massachusetts Amherst, Amherst MA
Investigators
Abstract
Emerging web applications are increasingly likely to use dynamic web data. Traditionally, requests for dynamic data---both time-varying and dynamically-generated---have been handled directly by servers and intermediate proxies haven't been allowed to process requests for such objects (web proxies have been primarily employed to disseminate data that is mostly static). A pure server-based approach for managing dynamic data is likely to limit the scalability of emerging web applications and increase their vulnerability to server failures. To overcome this drawback, novel proxy-based techniques will be developed to manage and disseminate dynamic web data. The proposed research will address two key issues: (i) data dissemination, which addresses the issue of disseminating time-varying web data using proxy-based push and pull techniques, and (ii) computation dissemination, which addresses the issue of moving computations from servers to proxies so as to dynamically generate web objects at a proxy. Both techniques have the potential of radically changing the way web proxies are designed and used. These techniques will need to satisfy three key requirements: user-cognizance (i.e., awareness of user and application requirements), intelligence (i.e., the ability to dynamically choose the most efficient set of mechanisms to service each application), and adaptivity (i.e., the ability to react to changing load characteristics). The proposed techniques will be evaluated using an eclectic mix of simulation, analysis and prototype implementation.
View original record on NSF Award Search →