With the progress of society and the development of science and technology, the data management of archives in colleges and universities has also changed from traditional manual operation to information technology system operation, which has taken a big step forward in the speed of data archive management. At the same time, following in the footsteps of the party and the state, according to the education reform measures issued by them, further implementation and implementation. With the innovation of education model, the reform of management method, and the expansion of teaching scale, the types and quantity of files are also increasing, and the scope is also expanding, which leads to the continuous increase of management difficulty. This requires data management to achieve informatization to meet the needs of society. The purpose of this paper is to study the optimization and upgrading of the special database management system of the archives of colleges and universities based on cloud computing. Based on cloud computing, this paper proposes the establishment of a special database management system for college archives and the methods and key technologies used in the process of optimization and upgrading. First, analyze the characteristics of the database management system, and then conduct demand analysis and functional analysis of the system. Secondly, research the advantages and disadvantages between the traditional data management of colleges and universities and the current latest database management system. Finally, according to the needs of the special database and system function requirements, the overall design of the archives management system is carried out, database modeling tools are used to optimize the database, and then its operation is studied, and system testing is carried out to enable it to meet the needs of university archives management. The experimental results show that the focus of database function optimization in different universities is different. University A prefers timeliness, accounting for 46%. University B thinks security is the most important, accounting for as high as 51%. University C is relatively even, both at 30%. The conclusion is to select the most practical and demanding functions for further optimization, and choose an overall design suitable for the special data of colleges and universities.