| 摘要 |
PROBLEM TO BE SOLVED: To distribute the loads so that a request for total delay consists of network delay and processing delay is satisfied among all edge devices and servers, even if the network delay is generally different between each edge device and each server.SOLUTION: In a load distribution control method used in a load distribution control system where a plurality of edge devices 3 for aggregating service request messages from a plurality of clients 1 and distributing them to servers 6, a plurality of servers 6 for processing the service request, and a controller 4 are connected by a network, each edge device 3 notifies the controller 4 of the information about the network delay observed between each edge device 3 and each server 6, the controller 4 determines the maximum allowable occurrence rate of the server 6 by using the worst value of network delay from each edge device 3 to some server 6, and notifies each edge device 3 of the worst value of network delay, and the edge device 3 determines coefficients for assigning a client 1 to each server 6, by using the maximum allowable occurrence rate of the server 6. |
