How Essbase treats sparsity issues
In Essbase, dimensions are partitioned into sparse and dense. A combination of members from the sparse dimensions defines (via sections) a subgrid in the cube called block.
Every block contains all the cells corresponding to all possible combinations of members from the dense dimensions. Only existing blocks are stored (in compressed form) to disk. An index keeps track of the stored blocks. Individual cells are addressed by means of offsets within the block.