Map reduce using coordination namespace hardware acceleration

A method for mapping and reducing data generated by a plurality of processes running at one or more distributed computing nodes sharing a coordination namespace, said method comprising:

• generating, by a local process running at a current node, a tuple corresponding to data generated by the local process, each said tuple having a tuple name and tuple data value corresponding to the generated data, said tuple generated by applying, by the local process, a mapping function to map the generated data to the tuple name and tuple data value;
• receiving, at a controller at the current node, one or more messages requesting a shuffleing operation for accumulating tuples associated with a specified tuple name;
• allocating, by the controller, responsive to a received shuffle message, a named data element corresponding to the specified tuple name in said coordination namespace at the current node or another node;
• storing, at one or more memory locations associated with said named data element, tuple data records including the tuple data values from each said one or more messages received at said current node or another node corresponding to data generated from said local processes specifying said tuple name;
• receiving, at the controller, a reduce message from a requesting process running at the current node or another node, said reduce message specifying the tuple name and an aggregation function;
• and
• returning, using said controller, a data corresponding to said aggregation function associated with the stored tuple data values associated with said tuple name.