At the present time one of the greatest problems of any computer system it is the obtaining of the data with the greater possible speed, avoiding times of delay for the users of these applications. By such reason, the administrators for systems always have looked for the form but adapted to optimize the applications, beginning by the data base, which one centralizes the data to be compiled. The importance of this is increased when having greater volume of stored information. It is as well as it is born the necessity to be able to make tasks of distribution of load between the devices that operate on this repository, as they are the processors, memory and hard disks. On this last device we will focus our study, since he is the one that grows with greater speed, having required but space for the data and is not single sufficient to acquire greater units of storage or to enclose them to already installed, but to look for the best form to balance the service load between these units. Frequently, database administrators use manual methods (statistical) or very basic algorithms (the first best one) to make this work, becoming but complex when being increased the number of objects of amount and the data base of storage units. This paper presents a GRASP algorithm for solve the GAP problem.