Line Item dimension:
When a dimension is having a character which is having higly unique values and large number of values, in this scenario we make that the characteristic as a line item dimension, then the sid of the characteristic will be equal ti dim id ,so there is no requirement of dim id as it is equal to sid,then the sys eliminates dim id.That means it doesnt have dim id.In fact table also only sid will be there and finally we are elimination the space required for dim id..
This means that the dimension contains just one characteristic. This indicator shows that no dimension table will be created; the SID table for the characteristic assumes the role of the dimension table. Omitting the dimension table has the following advantages:
• When loading transaction data, IDs do not have to be generated in the dimension table for the entries. If you do have a degenerate dimension, this number range operation can negatively affect performance.
• In the InfoCube star schema one table can be omitted. Potentially this table can be very large, where cardinality is high. This means that SQL-based queries become easier. In many cases, the database optimizer can select better run schedules.
However, this also has one disadvantage: You cannot include additional characteristics in a dimension that is marked as a line item at a later date. This is only possible with normal dimensions.
• When loading transaction data, IDs do not have to be generated in the dimension table for the entries. If you do have a degenerate dimension, this number range operation can negatively affect performance.
• In the InfoCube star schema one table can be omitted. Potentially this table can be very large, where cardinality is high. This means that SQL-based queries become easier. In many cases, the database optimizer can select better run schedules.
However, this also has one disadvantage: You cannot include additional characteristics in a dimension that is marked as a line item at a later date. This is only possible with normal dimensions.
High Cardinality
If a character is having highly unique values and more no of records we can declare it as cardinality height in the dimension properties. By default in bw sys all the dimension tables will have bit map indexes, if bcoz of high cardinality attribute the sys will create b-tree index.so b-tree indexes have better performance when we have bit map index. The cardinality can be checked in Transaction RSRV
If a character is having highly unique values and more no of records we can declare it as cardinality height in the dimension properties. By default in bw sys all the dimension tables will have bit map indexes, if bcoz of high cardinality attribute the sys will create b-tree index.so b-tree indexes have better performance when we have bit map index. The cardinality can be checked in Transaction RSRV
Example of High Cardinality:
As an Example: Billing Document Sales Cube:
This is when the Number of Entries in a DIM table exceeds 20% (or so) of the number of entries in the Fact Table:
using trans: LISTSCHEMA
Please note:- Can''t seem to get the Table working correctly.. assume the / as the column spacer:
Dimension / Entries / % Cardinality of Fact Table Entries.
DataPackage / 1,590 / 0.0%
Time / 3,170 / 0.1%
Unit / 393 / 0.0%
Billing Doc / 2,099,721 / 61.7%
Customer / 91,741 / 2.7%
Product / 14,345 / 0.4%
Organisation / 20,804 / 0.6%
Sales Doc / 1,922,274 / 56.5%
Equipment / 48,119 / 1.4%
Internal / 4 / 0.0%
Asset / 18 / 0.0%
Fact Table: / FZ_SALES / 3,402,323
This is when the Number of Entries in a DIM table exceeds 20% (or so) of the number of entries in the Fact Table:
using trans: LISTSCHEMA
Please note:- Can''t seem to get the Table working correctly.. assume the / as the column spacer:
Dimension / Entries / % Cardinality of Fact Table Entries.
DataPackage / 1,590 / 0.0%
Time / 3,170 / 0.1%
Unit / 393 / 0.0%
Billing Doc / 2,099,721 / 61.7%
Customer / 91,741 / 2.7%
Product / 14,345 / 0.4%
Organisation / 20,804 / 0.6%
Sales Doc / 1,922,274 / 56.5%
Equipment / 48,119 / 1.4%
Internal / 4 / 0.0%
Asset / 18 / 0.0%
Fact Table: / FZ_SALES / 3,402,323