Networking (SDN) is a new paradigm for configuring, controlling and managing computer networks. In SDN's logically centralized approach to network control, a reliable and accurate view of the network topology is absolutely essential. Most SDN controllers use a de-facto standard topology discovery mechanism based on OpenFlow to identify active links in the network. This paper evaluates the security, or rather lack thereof, of the current SDN topology discovery mechanism. We discuss and demonstrate its vulnerability to a simple link spoofing attack, which allows an attacker to poison the topology view of the controller. The feasibility of the attack is verified and demonstrated via experiments, and its impact on higher layer services is evaluated, via the example of shortest path routing. The paper finally discusses countermeasures, and implements and evaluates the most promising one.