Online hosts and networks are easy targets of network attacks due to their static nature, which creates an information asym- metry and makes them easy to attack and hard to defend. To break the asymmetry, Moving Target Defense was proposed to bring uncertainties to computer systems. It can be applied to all levels of protections, cov- ering applications, system software, operating systems, and networks. In this paper, we present, Sniffer Reflector, a new method to practice Moving Target Defense against network reconnaissance, which is usually considered as the very first step of most attacks. Sniffer Reflector employs Software-Defined Networking to disturb network reconnaissance. We use virtualization to provide an obfuscated reconnaissance result for attack- ers. Our method can be easily combined with existing security tools for network forensics as well. We have developed a prototype in a virtual local area network. Our experiment results show that Sniffer Reflector is effective and efficient in blurring various network reconnaissance.