System and Method for Implementing NUMA-Aware Statistics Counters

A system comprising:

• a plurality of computing nodes, wherein each computing node comprises one or more processor cores that collectively support multithreading and a memory storing program instructions that when executed on the one or more computing nodes cause the one or more computing nodes to perform:
• beginning execution of a multithreaded application that comprises one or more operations to increment a shared counter, wherein the shared counter is implemented as a data structure that is accessible by a plurality of threads of the multithreaded application that are executing on the plurality of computing nodes, wherein the data structure comprises a count value portion and a node identifier portion;
• initiating, by a given thread of the application, an increment operation targeting the shared counter;
• and
• determining whether to attempt to perform the increment operation or delay performance of the increment operation, wherein said determining is dependent, at least in part, on a value stored in the node identifier portion of the data structure;
• wherein the value stored in the node identifier portion of the data structure indicates whether one or more threads on a particular one of the plurality of computing nodes have priority to increment the shared counter or have requested priority to increment the shared counter.

A non-transitory, computer readable storage medium storing program instructions that when executed on one or more computers cause the one or more computers to perform:

• beginning execution of a multithreaded application that comprises one or more operations to increment a shared counter, wherein the shared counter is implemented as a data structure that is accessible by a plurality of threads of the multithreaded application that are executing on a plurality of computing nodes, wherein the data structure comprises a count value portion and a node identifier portion;
• initiating, by a given thread of the application, an increment operation targeting the shared counter;
• and
• determining whether to attempt to perform the increment operation or delay performance of the increment operation, wherein said determining is dependent, at least in part, on a value stored in the node identifier portion of the data structure;
• wherein the value stored in the node identifier portion of the data structure indicates whether one or more threads on a particular one of the plurality of computing nodes have priority to increment the shared counter or have requested priority to increment the shared counter.