The computational grid offers services for efficiently scheduling jobs on the grid, but for grid-enabled applications where data handling is a most relevant part in performance, the data grid kicks in. It typically builds on the concept of files, sites and file transfers between sites using services such as GRID-ftp, plus a Replica Manager to keep track of where replicas are located. Usually, careful original data layout is not considered because the data can be moved based on some replica optimization models to computing sites. We consider cooperative shared computations for huge data that is distributed over a variable number of sites. In this context, data layout is a relevant problem because moving very large amounts of data is too expensive. Consequently, it is a balance between data layout and load-balancing capabilities on top of the basic infra-structure that is the most crucial issue. In this paper we propose a mixed replication layout and load-balancing approach to yield very good performance results in such environments.