Abstract: Systems, methods, and computer program products are disclosed for dynamically determining an optimal node of an overlay network for client attachment. An optimal node can be selected from a group of potential attachment nodes in the overlay network through which a client can establish a connection to the overlay network. Network parameter training data can be determined and used to train a network optimization model. Once trained, the network optimization model can be used to determine—based on network parameter data associated with the group of potential attachment nodes—the optimal node for the client to attach to. If a different node than a current attachment node is determined to be a new optimal node, a connection path can be established between this new optimal node and the client, and an existing connection path between the client and the previous optimal node can be terminated or simultaneously maintained.

Abstract: Systems, methods, and computer program products are disclosed for dynamically determining an optimal node of an overlay network for client attachment. An optimal node can be selected from a group of potential attachment nodes in the overlay network through which a client can establish a connection to the overlay network. Network parameter training data can be determined and used to train a network optimization model. Once trained, the network optimization model can be used to determine—based on network parameter data associated with the group of potential attachment nodes—the optimal node for the client to attach to. If a different node than a current attachment node is determined to be a new optimal node, a connection path can be established between this new optimal node and the client, and an existing connection path between the client and the previous optimal node can be terminated or simultaneously maintained.