Manufacturing systems are crucial for the Industry. A frequent objective is to improve the throughput of the system, which can be done by carefully adding or removing buffers in the system or by improving the machine performances. The Toyota Central Research Laboratories have recently developed a number of simulation based methods to understand and predict the behavior of a manufacturing system. These methods include a reliable quantitative bottleneck detection, resulting in a machine sensitivity analysis including a prediction model of the effect of machine changes, and a blocking and starving analysis, resulting in a prediction model of the effects of buffers and a subsequent buffer optimization. The novel idea of these methods is a holistic view of the manufacturing system, i.e. understanding the system by analyzing the relations between the machines instead of analyzing machines independently. This allows a much better understanding of the manufacturing system, and enables the optimization of the manufacturing system using only a single simulation. These methods are being evaluated at selected companies of the Toyota group, and have also generated great interest in academia and industry. This paper provides a framework of the holistic manufacturing analysis and a summary of the developed methods.