This document compares the performance of Apache Spark and Apache Flink for stream processing tasks. It develops two evaluation workloads using aircraft trajectory data from the DatAcron project: 1) computing statistics for each trajectory, and 2) detecting changes in aircraft sectors. The document finds that Flink has lower latency than Spark Streaming due to its stateful, low-latency processing model. However, Spark Streaming can achieve higher throughput than Flink by increasing batch durations. Overall, the document concludes that Flink is generally better suited than Spark for stream processing workloads.