Grid computing represents a new challenge for enterprises looking for computing and storage resources. Business grid is the new frontier that, with the wide adoption of the notion of services, is paving the way through a real shift of the grid computing paradigm from the scientific to the commercial sector. This imposes among the others the solution of security issues, as well as the development of reliable mechanisms to access data. In this paper we address the problem of data fault tolerance in Grid through the adoption of the redundant residue number system. From each file to be stored, a set of n elements is generated which are allocated to n + r different containers. If a disrupting event occurs, the r new elements will allow a correct reconstruction of the original file. The user will autonomously select both the fragmentation number and the degree of replication, that best fit his/her needs.