The techniques of data store and presentation in SAP

 

The concepts dealing with how data is stored inclue transparent tables, database clusters, pooled tables and the concept dealing with data presentation is logical database.

Transparent tables: Transparent tables are abap dictionary tables that have a one-to-one relationship with database tables stored in the underlying database management system.

Database views: Theses are views in the ABAP dictionary that joints tables for data selection purpose.

Pooled tables: In earlier days, SAP R/3 needed more tables than the underlying database management system would allow and had to group tables together in pooled tables. In other words, table definitions are in SAP and behave as transparent tables, but on the database no single corresponding physical table exists. Technically, these pooled tables had the table name as its key, and the data would be saved as a long raw string in a field in the database. When a pooled table is read, the data is parsed back into the data structure of the SAP table it is supposed to represent.

Cluster tables: SAP uses cluster tables, which are similar to pooled tables. The difference is that in cluster tables, complex data objects in program memory can be saved in a cluster without any flattening operations(as would be necessary if this data had to be committed to a transparent table). Technically, data in clusters is saved as raw strings. Because of requisite parsing and fattening transformations needed to read these tables, performance can be an issue. Many HR payroll tables are defined as cluster tables.

Logical databases. This is a hierarchy of reporting structures that populated with values via an ABAP program. The structures in the hierarchy may or may not represent actual tables. Logical databases are popular for reporting because developers who program against its structures do not have to invest any development time into designing the data retrieval to the actual tables. In R/3, logical databases are a popular reporting structure for classic infosets.

Data cluster : a data cluster is a grouping of several data objects. Elementary fields, field strings and internal tables can be grouped in a data cluster.