An automated method for provisioning a grid used to run a load test on a target website includes sending one or more requests in a multi-threaded manner to at least one cloud provider, the one or more requests for an allocation of N load server instances and M result server instances which comprise the grid. Requests received back from the cloud provider are also handled in a multi-threaded manner; any errors occurring during the allocation being corrected automatically. The N load server instances and the M result server instances are then verified to be operational and correctly running software deployed to provide defined test services. Errors identified during the verification are automatically corrected either by attempting to restart a failed instance or allocating a different instance.