Systems and methods for intelligent load balancing of hosted sessions

A system, comprising:

• a computing device comprising a processor coupled to memory and a network interface, the one or more processors configured to:
• determine a plurality of metrics for each of a plurality of machines configured to connect computers with hosted sessions;
• receive, from a client device, a request to establish a connection with one of the plurality of machines to access a hosted session;
• determine a score for each of the plurality of machines based at least on the plurality of metrics for each of the plurality of machines;
• select an machine from the plurality of machines as a function of the score and a resource cost of the machine;
• and
• cause the client device to connect to the selected machine for the hosted session;
• assign resource costs of zero to a subset of the plurality of machines responsive to each machine of the subset currently hosting a hosted session;
• and
• calculate a cost for each machine of the plurality of machines other than the subset assigned the resource costs of zero based on machine specifications of computing devices executing each machine of the plurality of machines other than the subset.

A method, comprising:

• determining, by one or more processors processor, a plurality of metrics for each of a plurality of machines configured to connect client devices with hosted sessions;
• receiving, by the one or more processors from a client device, a request to establish a connection with one of the plurality of machines to access a hosted session;
• determining, by the one or more processors, a score for each of the plurality of machines based on the plurality of metrics for each of the plurality of machines;
• selecting, by the one or more processors, an machine from the plurality of machines as a function of the score for the machine;
• and
• causing, by the one or more processors, the client device to connect to the selected machine for the hosted session;
• concatenating, by the one or more processors, the plurality of metrics, an account identifier of an account accessed at the client device that transmitted the request, and an application identifier of a first application executing on the client device that transmitted the request to generate a feature vector;
• and
• executing, by the one or more processors, a machine learning model using the feature vector as input to determine the score for each of the plurality of machines.