Methods and systems to limit the duration of a service interruption caused by a failed middleware application server are disclosed. One example method and system includes monitoring the operational status of a first virtual machine operating as a primary server and the operational status of a second virtual machine operating as a stand-by server and, based on the monitored operational status of the first and second virtual machines, performing a hot-swap to cause the second virtual machine to operate as the primary server and the first virtual machine to operate as the stand-by server. The first and second virtual machines are implemented on a same host processing system. After the hot-swap, the first virtual machine is restarted. In some examples, the first virtual machine and the second virtual machine are implemented on a same host processing system. Performing the hot-swap can include causing the first virtual machine to be uncoupled from a network and causing the second virtual machine to be coupled to the network. In some examples, the method also includes causing copy of a first file system used by the primary server to be stored as a second file system for use by the stand-by server.