A system and method for reserving resources for a virtual machine to use during failure of an underlying node are disclosed. The system and method include a scheduling module that is configured to reserve resources for the virtual machine on at least one candidate node. To reserve the resources, the scheduling module is configured to identify a list of candidate nodes based upon a qualifier function, rank the candidate nodes based upon a priority function, and determine the candidate nodes that satisfy a high availability threshold. The scheduling module is also configured to select a highest-ranked candidate node that satisfies the high availability threshold and reserve the resources on the highest-ranked candidate node.