A data stream's order describes how the rows in that data stream are arranged. An order is characterized by a list of column names, their data types, whether each of the columns is arranged in an ascending or descending order, and whether NULL values are ordered before or after other values.
When considering the order of a data stream, you should keep in mind its distribution. For example, a data stream distributed on column 'a' indicates that the data stream is distributed across vWorkers by hashing the value of column 'a'. If that data stream is ordered on column 'b' on each vWorker, the data stream is said to be ordered on the columns (a, b).