The talk shows how to use Spark along with other open source technologies to support IoT and Smarter Cities use cases like traffic monitoring on Bluemix. The work was developed by IBM, ATOS and the University of Surrey for the EMT Madrid Traffic bus company in the context of COSMOS, a European project. It performs machine learning on historical data to generate rules to respond in real-time to live data streams. Data is continuously collected from external web services and data sources, indexed and stored using Apache Kafka, Pinterest Secor with specialized extensions, OpenStack Swift and Node-RED. Historical data is analyzed with a newly developed Spark SQL driver that exploits the indexes generated and significantly improves performance.