Your customer wants to stream large amounts of real-time data from multiple sources at the application level into their Apache Spark platform for transformation before it goes into their machine learning platform.
In this scenario, which technology should be used for this task?