Data streaming systems are a computerized system that is built with the purpose of managing and processing the data in motion. The sheer size, variety, and velocity of big data add further challenges to these systems. Such systems are designed to manage relatively simple computations, such as one record at a time or a set of objects in a short time window of the most recent data:

In a data-steaming system, the computations are done in near real-time, sometimes in memory, and as independent computations. The processing components often subscribe to a system, or a stream source, non-interactively. This means they send nothing back to the source and nor do they establish interactions with the source. The concept of dynamic steering involves dynamically changing the next steps or direction of an application through a continuous computational process using streaming. Dynamic steering is often a part of data streaming management and processing. A self-driving car is a perfect example of a dynamic steering application, but all data streaming applications fall into this category, such as the online gaming. Amazon Kinesis and other open source Apache projects, such as Storm, Flink, Spark Streaming, and Samza, are examples of big data streaming systems. Many other companies also provide streaming systems for big data, which are frequently updated in response to the rapidly changing nature of these technologies.