Dimensions in Hyperion Essbase

Types of Dimensions:

Basically dimensions are two types.
1. Standard Dimensions
2. Attribute Dimensions

These are further classified into sparse and dense dimensions.
Standard dimensions may be sparse or dense but attribute dimensions are always dense in nature.

Sparse Dimensions:
The selection of sparse dimensions depends upon the business. If data is not available for majority of the member combinations that dimension will keep as sparse dimension. For every unique member combination an index will get formed and a block will get formed for the same combination. That means every sparse dimension maintains one index file and one page file (having data).
Number of indexes = Number of blocks = Multiply the number of members from one sparse with the number of members from the other sparse dimension and so on.

Ex: Take 2 sparse dimensions
Products (sparse) - 3 members
Market (sparse) - 4 members
No of indexes=3*4 =12 = no of blocks

Disadvantage of Sparse dimension:
When all dimensions are sparse, for every sparse combination it will create indexes. So it will take more time to search the index itself. So don’t go for all dimensions are sparse.

Dense Dimensions:
Data exists for majority of the member combinations that dimension will be dense. Dense dimensions affect the size of the block,
Number of cells = multiply the number of members from one dense dimension with number of members from the other dense dimension.
Block size = Number of cells * 8
Ex: Time (dense) - 4 members
Measures (dense) - 3 members
No of cells = 4*3 =12
Block size = 12* = 96

Note:
1. Ideal block size 8k to 100k
2. If the block size is between ideal block sizes we can make that dimension as sparse.
3. With sparse and dense dimensions we can reduce the load and calculation times.

Attribute dimensions:
Generally attribute dimensions are sparse in nature. Attribute dimensions are just like as shared members. Once we create the attribute dimension the member will not having any consolidation operator. The attribute dimension must be at the end of the outline. The attribute dimension can be associated to only base dimension. Not for other dimensions.

NOTE: For one base dimension we can have any number of attribute dimensions.