Lessons Learned from Integrating Batch and Stream Processing using IoT Data

The unbounded data streams generated by IoT sensors/devices are posing many technical challenges and requires a one-size-fits-all solution to cope with the massive amount and the high speed of the incoming IoT data arriving simultaneously. In this study, we try to integrate batch and stream processing in a unique system as a premise to handle Volume and Velocity aspects of IoT data simultaneously. In order to handle current, outdated, and historical IoT data streams, we built a cloud architecture to execute the analytical workflows using both batch and stream processing in a synergetic manner. A smart parking case study is used to evaluate the architecture and two experiments are implemented to demonstrate a web application for predicting parking spot availability. Herein, we learned our lessons that there are several hindrances to finding a middle ground where current, outdated and historical IoT data streams can be used in a strategic way.