A method, apparatus, and computing node for optimizing function call chain execution.

Abstract

A function call chain execution optimization method is applied to a first compute node in a computing cluster. The method includes: generating a data dependency graph based on the function call chain to be executed during function runtime. This data dependency graph records data access operations and computation operations in the function call chain, as well as the data dependencies between these operations; selecting a second compute node from the computing cluster based on the data access operations recorded in the data dependency graph, wherein the second compute node is accessed more frequently than at least one other compute node in the cluster; and scheduling the data dependency graph to the second compute node so that the second compute node executes the function call chain based on the data dependency graph. For example, the computing cluster can be, but is not limited to, a serverless computing platform. This method can reduce remote data access and function calls during the execution of the function call chain and improve the accuracy of data affinity scheduling.

Images