The 35th IPP Symposium

Infrastructures for Stream Processing and Dissemination

Ugur Cetintemel, Brown University

Data streams are proliferating - they are generated by physical sensors (cameras, climate sensors, telescopes, GPS), cyber sensors (network monitors), computational processes (scientific simulations), and humans (blogs, RSS). Many existing and newly emerging applications require sophisticated, timely processing and distribution of this deluge of data. However, existing systems software fall short of meeting the scalability, availability, and performance requirements of these applications, thereby requiring new generic infrastructures.

This talk will provide an overview of our recent work on two such software infrastructures. We will first present the high-level design and key features of Borealis, a scalable distributed stream processing engine that facilitates low-latency processing of high-volume data streams. We will then provide an overview of XPORT, an adaptive profile-driven stream dissemination system. Both systems significantly ease the development and deployment of robust and efficient stream-based applications.