Alternatively, to avoid a memory overflow, we may employ a queue that may ignore incoming messages if it is full. The following marble diagram depicts a queue that has a size of 2 elements and is characterized by dropping elements on overflow (Diagram 3.5):

In general, this technique respects the resources' limitations and makes it possible to configure the capacity of the queue based on the resources' capacities. In turn, embracing this kind of queue is a common practice when the importance of the message is low. An example of a business case may be a stream of dataset-changed events. In turn, each event triggers some statistical recalculation which uses the entire dataset aggregation and takes a significant amount of time in comparison with the incoming events quantity. In that case, the only important thing is the fact that the data set changed; it is not vital to known which data has been affected.