Child pages
  • BWonHANA: Large Data Volumes and Partitioning
Skip to end of metadata
Go to start of metadata

Introduction/Overview

In order to avoid performance issues, as a general rule of thumb, it is recommended to restrict the data volume in BW Providers as an ADSO to fewer than 1 billion records per physical database table. Depending on the type of the BW Provider, there are the following two important consulting notes:

  • ADSO: 2374652  Handling very large data volumes with advanced DataStore objects in SAP BW on SAP HANA and BW/4HANA
  • DSO/InfoCube: 2019973  BWonHANA - Handling BIG data tables

Basically, there are two ways to avoid too big BW Provider database tables: 

Remark to Semantic Groups: in contrast to Semantically Partitioned Objects  (SPOs), the Semantic Group creates advanced DataStore objects that can be edited individually and it is even possible to include existing advanced DataStore objects in a semantic group. For more information about semantic groups, see the SAP Help Portal.

If very big BW Providers cannot be avoided, you need to take into consideration that there is a limit of 2 billion rows for a Hana database partition(see note above). So, in order to avoid issues with this limit, you can use the possibility to create database partitions for an BW Provider in order to distribute the data over database partitions (on first or second level) which can also be stored on different nodes. Basically, there are two different ways of creating Hana db partitions

  • 'Automatic DB Partitioning'(see below)
  • User Defined DB Partitioning(see below)

Database Partitions of BW Provider Tables

Automatic Partitioning

Check point 17 of

note 2044468  FAQ: SAP HANA Partitioning 

where you can find a detailed description on how tables of BW Objects are partitioned. Basically, BW creates all partitions during the first activation of the BW Provider(based on the table placement configuration, see notes below). A manual intervention is only required if you already know that a certain Provider will have to handle a huge data volume in the near future, see note 2069235. In case the provider reaches a big size after some time, it is important to know that partitions are not adapted dynamically(exception: see note 2081135) . If an adaption is required, a landscape redistribution must be executed to adapt the number of partitions - see

  • 1908075 BWonHANA: Table placement and Landscape Redistribution
  • 2334091 BW/4HANA: Table Placement and Landscape Redistribution.

This kind of automatic partitioning(Hash for ADSOs and Roundrobin for InfoCubes) is mainly used to distribute the data to various nodes in the SAP HANA database.

See also:

  • ADSO: 2374652  Handling very large data volumes with advanced DataStore objects in SAP BW on SAP HANA and BW/4HANA
  • DSO/InfoCube: 2019973  BWonHANA - Handling BIG data tables

User Defined Partitioning

For an ADSO it is possible to define partitions when creating this object in the BW Modeling Tools the following way:

 

Second-Level Partitions are created for the active table of the ADSO.

See also:

SAP Online Documenation

Semantic Groups (BW/4Hana)

SAP Notes

  • 2374652  Handling very large data volumes with advanced DataStore objects in SAP BW on SAP HANA and BW/4HANA
  • 2019973  BWonHANA - Handling BIG data tables
  • 2143736  FAQ: SAP HANA Table Distribution for BW
  • 2044468  FAQ: SAP HANA Partitioning
  • 1908075  BWonSAP HANA: Table placement and landscape redistribution
  • 2334091  BW/4HANA: Table Placement and Landscape Redistribution
  • 2143736  FAQ: SAP HANA Table Distribution for BW
  • 2081135  Enable "Dynamic Range Partitioning" feature for DataStore objects (advanced)
  • 2069235  How to Change the Initial Number of First-level Partitions for BW generated Tables in SAP HANA
  • No labels