Microsoft's Machine Translation system has been publically available since September, 2007. So what does it take to put together an environment capable of translating text in 26 languages pairs at a rate of more than 1,000,000 wpm? We present some of the issues that we faced in preparing to go live, such as architectural design dilemmas, security concerns, adequate redundancy to assure 100% uptime in the data center, integrating a 3rd party application into our framework, hardware choices, etc. Our talk focuses on the lessons learned during this process and highlights some of the issues that are otherwise easily forgotten when attempting to transfer a research project to a production environment.