The topic of this article is to review the analysis method of the high frequency financial data in recent two decades. The high frequency financial data is a new data type which the price is recorded as soon as the transaction occurred. Therefore, this kind of data is usually constructed by using the stochastic volatility model. How to estimate the integrated volatility is the key point of the research. A further problem is, when the observations which are contaminated by the microstructure noise, how to estimate the integrated volatility. On the other hand, in this article, the research on the goodness-of-fit of the distribution of the volatility process is also introduced. Finally, the extension problems are also provided.