LoadMaster by default performs 'connection-based load balancing' – meaning that although many http requests are pipelined into a single connection, load balancing is performed using the information associated with the first request in the connection only.

Some customers with specific application demands require instead 'request-based load balancing" where each request in a connection is load balanced separately from other requests.

HTTP request-based load balancing is disabled by default and can be enabled using the Reschedule on every HTTP Request check box in a Virtual Service's Advanced Properties. This can be used in combination with any of the Selection Methods supported under Standard Options, but should not be used with HTTP/2 workloads.