SYSTEM AND METHOD OF FUNCION PLACEMENT CONTROL BASED ON REINFORCEMENT LEARNING IN HIERARCHICAL DISTRIBUTED COMPUTING ENVIRONMENT

摘要

When multiple computing nodes execute a function requested by applications of terminals in an environment hierarchically distributed in a network, the function uses reinforcement learning to execute the function in the optimal computing node while guaranteeing the execution completion time limit condition as much as possible. A system and method for autonomously controlling the execution location of A computing node according to an embodiment of the present invention receives a function execution request from a terminal, determines whether to accept the request, and receives a token bucket size update message for each function type from a resource allocation controller, a function arrangement controller, a function type A computing cluster scheduler that sequentially allocates computing resources to the function execution requests using a separate waiting queue and receives a computing resource allocation update message from the resource allocation controller and executes the function, and executes the function when the function execution is completed It may include a computing resource cluster for transmitting the result to the terminal.