Taking Stream Processing out of its Ivory Tower: Stream Processing in the Age of IoT and Proliferation of Video Content
13 February 2017
The emergence of Internet of Things and the proliferation of video content has unlocked a slew of new streaming applications. In this paper we analyze a diverse set of such applications and distill their requirements into separate groups, most notably: environment-awareness (i.e. actuation), dataflow dynamism, multi-modal processing, and geographical distribution (i.e. wide-area and edge). We argue that the state-of-the-art stream processing platforms are ill-equipped to meet these requirement and introduce World-Wide Streams (WWS), a novel stream processing platform that has been designed with those requirements in mind. WWS has many innovative concepts and features. Of these, we highlight a few in this paper. More specifically, we introduce our flow language for authoring streaming applications, our interaction mechanism between WWS and its streaming sources as well as our novel solution for dataflow dynamism. The WWS platform has been used to build and showcase a number of different streaming application prototypes in the last few years
