In s-Server nomenclature, a pipeline refers to a set of sources, internal streams, views, pumps, and sinks. You run analytics inside pipelines, using SQL, UDXs, UDFs, machine learning, and so on.

A typical pipeline begins with a source, proceeds through a number of s-Server internal streams and views, and ends in a sink. Pumps (pairs of select/inserts) move data through the pipeline.