Compile-time grouping of tuples in a streaming application

摘要

A system and a method for initializing a streaming application are disclosed. The method may include initializing a streaming application for execution on one or more compute nodes which are adapted to execute one or more stream operators. The method may, during a compiling of code, identify whether a processing condition exists at a first stream operator of a plurality of stream operators. The method may add a grouping condition to a second stream operator of the plurality of stream operators if the processing condition exists. The method may provide for the second stream operator to group tuples for sending to the first stream operator.

主权项

1. A method for initializing a streaming application for execution on one or more compute nodes with each compute node adapted to execute one or more stream operators, comprising:
examining, with a compiler system, an operator graph with a plurality of stream operators having a first stream operator written in a first set of programming instructions that outputs a plurality of tuples to a second stream operator; determining, with the compiler system, an existence of a defined processing condition at the second stream operator of the plurality of stream operators during compile-time, the defined processing condition includes a tumbling windowing condition that arranges a subset of tuples to be processed by the second stream operator from an input of the plurality of tuples by storing incoming tuples until a window is full, and flushing the stored tuples from the window; and modifying the first stream operator, with the compiler system during compile-time, by compiling code that results in adding a grouping condition into the first stream operator of the plurality of stream operators in response to the determining of the existence of the defined processing condition at the second stream operator, wherein the grouping condition causes the first stream operator to output a specific group of tuples from the plurality of tuples to the second stream operator, wherein adding a grouping condition causes the first stream operator to have a second set of programming instructions different than the first set of programming instructions.