Systems and methods herein save one or more parameters related to NVMe® data in one or more memory mapped files or in shared memory region(s). When a storage target experiences an interrupt event (e.g., loss of connectivity, system crash, application crash, etc.), the storage target may be restarted (e.g., a new instance formed in the case of a software-defined storage target) that uses the stored parameter data to recreate the controller context (e.g., NVMe® controller and TCP (transmission control protocol) connection), and the storage target may resend the last TCP acknowledgement, thereby causing the host to send or resent any NVMe/TCP data that follows the last successfully received NVMe/TCP data—thus, any data that the host sent during the storage target down time, which may have been lost, is resent. Embodiments keep a host unaware of a storage target interrupt and remove the need for TCP reestablishment.