1 Real Time Streaming Data Grid Applications Geoffrey C. Fox, Mehmet S. Aktas, Galip Aydin, Hasan Bulut, Shrideep Pal- lickara, Marlon Pierce, Ahmet Sayar, Wenjun Wu, and Gang Zhai Community Grids Lab, Indiana University 501 North Morton Street, Suite 224 Bloomington, IN 47404 {gcf, maktas, gaydin, hbulut, spallick, mpierce, asayar, wewu, gzhai}@cs.indiana.edu Abstract. We review several aspects of building real-time streaming data Grid applications. Building on general purpose messaging system software (Na- radaBrokering) and generalized collaboration services (GlobalMMCS), we are developing a diverse set of interoperable capabilities. These include dynamic information systems for managing short-lived collaborative service collections (“gaggles”), stream filters to support the integration of Geographical Informa- tion Systems services with data analysis applications, streaming video to sup- port collaborative geospatial maps with time-dependent data, and video stream playback and annotation services to enable scientific collaboration. 1 Introduction This paper describes research work of the Community Grids Laboratory on Grids built around streaming data sources. This work builds upon general purpose messag- ing middleware (NaradaBrokering [1, 2, 3]) and incorporates a diverse set of services that include audio/video conferencing (GlobalMMCS [4]) and Geographical Informa- tion System services [5]. The architecture and core services are summarized in a recent companion publication [6]. Here, we examine more closely applications and additional functionality that are being integrated into the overall system. A critical idea in our approach is to view both services and messages (and streams as ordered set of messages) as “first class” entities. The law of the millisecond [7] suggests one should use this type of message oriented middleware (software overlay network) when one can afford latencies of a millisecond or more. This is characteris- tic of all systems with significant geographic distribution and non-specialized inter- connect. As discussed in [6-8], NaradaBrokering is capable of supporting millisec- ond messaging in a diverse range of applications, ranging from binary data streams to XML-based Web Service messages. For scientific applications, the data deluge [8] suggests the growing importance of real time data assimilation with the integration of sensors, databases and simulation codes. Similarly, the geographically distributed nature of much current research requires collaboration tools. Our research thus focuses on the reuse of concepts and software as well as integration of data-driven and collaboration-driven real-time prob-