In Software-Defined Networking (SDN), network applications use the logically centralized network view provided by the controller to remotely orchestrate the network switches. To avoid the controller being a single point of failure, traditional fault-tolerance techniques are employed to guarantee availability, a fundamental requirement in production environments. Unfortunately, these techniques fall short of ensuring correct network behaviour under controller failures. The problem of these techniques is that they deal with only part of the problem: guaranteeing that application and controller state remains consistent between replicas. However, in an SDN the switches maintain hard state that must also be handled consistently. Fault-tolerant SDN must therefore include switch state into the problem. A recently proposed fault-tolerant controller platform, Ravana, solves this problem by extending fault-tolerant SDN control with mechanisms that guarantee control messages to be processed transactionally and exactly once, at both the controllers and the switches. These guarantees are given even in the face of controller and switch crashes. The elegance of this solution comes at a cost. Ravana requires switches to be modified and OpenFlow to be extended with hitherto unforeseen additions to the protocol. In face of this challenge we propose Rama, a fault-tolerant SDN controller platform that offers the same strong guarantees as Ravana without requiring modifications to switches or to the OpenFlow protocol. Experiments with our prototype implementation show the additional overhead to be modest, making Rama the first fault-tolerant SDN solution that can be immediately deployable.