Partitioning

Use

The database tables that are generated when you create new archive information structures can grow large over time with the use of these infostructures. As a result, the load on the database increases and the handling of the infostructures and tables becomes more difficult. To combat this effect, you can use partitioning to divide up the infostructures in such a way that the system distributes the data across several tables instead of writing it into one single table. This facilitates the administration of the database, which is more difficult with large data volumes.

Features

Configuration of Partitioning

You can call partitioning from the Archive Retrieval Configurator (transaction SARJ), by entering an infostructure and choosing Goto ® Partitioning.

To configure the partitioning function specify in the From Date field the date as of which the system is supposed to write the entries for the infostructure into a new table. The date refers to the creation date of the archiving session. In the Table Name field you can enter the name of the table that is to be generated. If you do not want to specify a name, the system automatically generates a table name that corresponds to the infostructure. The tables are not generated until the first data record is being written, which would be either during the delete phase or during the subsequent filling of the infostructure.

The following configuration data has been entered for partitioning:

From Date

Table Name

01.01.0001

02.01.2005

04.14.2005

Based on these entries, the system would generate a separate database table for the following time periods: Until 01.31.2005, 02.01.2005 until 04.13.2005 and as of 04.14.2005.

For a detailed example of a partitioning scenario, see Example: Partitioning of an Infostructure.

The configuration entries for partitioning logically belong to the infostructure. This means it is cross-client and is transported and shipped together with the infostructure. Changing the configuration settings means a modification of the infostructure. Therefore never change the partitioning of infostructures that are not part of your namespace. This particularly applies to the SAP infostructures (those shipped by SAP) in customer systems. For more information see Copying Infostructure Data.

The partitioning concept is so flexible that you do not have to specify that partitioning is to take place right at the beginning. With large data volumes rebuilding the infostructure would be very resource intensive. It would make sense, therefore, to leave already built data as is. All accesses to the Archive Information System remain intact. This applies to the Archive Explorer and to programmed accesses with the function module AS_API_READ or other interfaces to the Archive Information System.

Predefining Tables

In addition to partitioning you also have the option of predefining a table or several tables for a specific infostructure. This has the following advantages:

·        The name of the predefined table is determined at the time configuration takes place. You can choose a name from within your namespace - a name that matches your requirements. If you do not specify a table name the system automatically generates a name, which you cannot influence.

·        The table you specify is generated in the development or Customizing system and is then transported to the production system. This allows you to still make changes to the table, such as other technical settings or a different short text. In addition, the table has the same name in all systems, which is not the case for the automatically generated AS tables (ZARIX tables). This facilitates the administration of the system landscape.

When you specify a table in the configuration, the system checks whether this table already exists. If so, it checks whether the table matches the infostructure and possibly rejects the entry you made. If the table does not yet exist, the system sends a message asking whether or not you want to generate the table. If you say yes, the table is generated in such a way that it corresponds to the infostructure.

We recommend that you always let the system generate these tables. In this way you can be sure that the structure of the table always matches the infostructure. We also recommend that for the object directory entry you enter the same data as that of the infostructure. This is particularly important for package assignments.

Displaying Partitioning

You can use the display function to view the current settings of the partitioning of an infostructure. You can call up the display function from within the status management of the Archive Information System by choosing Goto ® Partitioning or from within the display of an infostructure in Customizing by choosing Technical Data.

·        Partitioning Configuration

This area on the display screen shows how partitioning was configured.

·        Current Partitioning

This area shows which tables the system used during which time periods so far. For the time periods that were entered in the configuration and that appear in this area, the system will use the assigned table also in the future. When the first record for the corresponding interval is written to the infostructure, the system sets the Change On date for that particular table. The entry that does not have a change date contains the table that was generated when the infostructure was first activated (normal AS table). This table only serves as a reference structure for partitioned infostructures.

Tables that have the G (Table Generated) indicator, have been generated.

If an entry only has the R (Read only) indicator, it means that the Archive Information System does not write any more entries to this table. It only reads from this table. This happens when you change partitioning, by, for example, grouping several time periods together. If this is the case, the system sets the tables used up until then for the individual tables to read only mode.

·        Assignment of Tables and Files

This area shows all the archive files that have entries in the infostructure. It lists the corresponding tables where the entries can be found. This relationship is unique. A table always contains the complete set of entries of an archive file. The system uses this information especially for the deletion of infostructure contents.

See also:

Example: Partitioning of an Infostructure