Data that is rarely used or not used any more at all in Data Warehouse processes or for analysis is known as non-active data. Despite this, this data should remain in the BW system. The concept for handling non-active data makes it possible to suppress non-active data from the main memory when memory bottlenecks occur and if required only to load the relevant, request-based partitions back into the main memory. This has a positive effect on hardware sizing when dealing with a large amount of non-active data.
In the Non-Active Data Monitor in the SAP HANA Database (transaction RSHDBMON) you can monitor which objects in your BW system are set for more rapid suppression. This is normally data in Persistent Staging Area tables and tables for write-optimized DataStore objects. The monitor also allows you to check that objects that are relevant are not suppressed so rapidly. The monitor provides an overview of all InfoCubes and DataStore objects (standard DSOs) as well as the Persistent Staging Area tables.
The release version of the SAP HANA database is SAP HANA Support Package Stack 05.
The concept of non-active data has been activated.
More information: Activating the Handling of Non-Active Data During Main Memory Bottlenecks and SAP Note 1776749 .
You are in the Non-Active Data Monitor in the SAP HANA Database (transaction RSHDBMON).
You can call also call the monitor from the Data Warehousing Workbench (transaction RSA1). Choose .
Initially, the upper part of the screen provides you with information about the system, while the lower part contains a table with information about the assigned and used memory consumption of the various services of the host.
Setting the search criteria for the detailed selection
To view information about the handling of non-active data belonging to a given object, choose . The Detailed Selection dialog box appears.
Restrict the search to the required object type. You can select from the following: InfoArea, InfoProvider, DataSource and the more runtime intensive options All Objects and All Objects with Unloads.
All selections flagged with Runtime Warning can entail very long runtimes, especially when run for the first time. Subsequently, these operations are faster, as the relevant accesses are buffered.
Enter the technical name of the object. You can use input help for this.
Define the search method:
Exact Match: The system searches for the exact technical name of the object of the specified object type entered in the fields described above.
Search Template: The system uses the entries made as a search template for the analysis.
Define which of the object's tables to search for. You have the following options:
Only tables with the setting "Early Unload"
Only tables without the setting "Early Unload"
If both options are selected, the system searches for all tables. This is the default setting.
Note that these settings function as a filter on all other windows.
Make other settings:
Unload Information in Header Area (Warning: Runtime): If you choose this option, the following unload information is displayed in the header area:
Total Unloads via Partitions since the time selected in the next step: Suppression is performed at column level. There are x partitions, of which suppressions are performed y times. The Total Unloads via Partitions indicates the maximum suppressed columns per partition.
Data has been suppressed once from partition 3 and three times from partition 5. Total Unloads via Partitions would be four in this case, as data has been suppressed from all partitions in the system four times.
Number of these that are unloads on reporting/staging partitions
Number of Tables Affected
Number of these that are reporting/staging tables
Unloads since: You can enter an appropriate point in time here.
No Refresh During Navigation: This setting is selected by default in order to keep the runtime short. It means however that you do not see changes made to the data during analysis.
If required, you can subsequently update the data display at any time by choosing .
Choose . In the upper screen area, the monitor shows you the unload information if you have selected this option. In the lower screen area it contains a table with detailed information about handling of non-active data in the SAP HANA database, filtered to the object set in your search query.
Analysis of Displayed Data
Depending on the search query, you see the tables of objects belonging to an InfoArea, InfoProvider or DataSource. By clicking on the underlined information you can navigate to further details:
Column InfoProvider: If you click on the technical name of an InfoProvider or a Data Source, the system displays the details for all tables participating in this object and all partitions.
Column Unloads: If you click on the number of unloads for a given object, the system displays the details for all affected partitions (partition ID, host, early unload, unloads, unload reason).
Note the following exception: If the system shows -1 as the partition ID, this means that the entire table has been suppressed from the main memory (for example with unload reason EXPLICIT, meaning manual unload).
Column Available in Memory?: : You have the following options:
Status of Availability in the Main Memory
The table is not available in the main memory.
The entire table is available in the main memory.
If you click on status FULL for a table, see the details such all entries for the memory, the number of records and the last access time,
Parts of the table are not available in the main memory. To be able to tell which one it is, you can click on the PARTIALLY status for the relevant table. You then see all columns of a partition.
If columns of a partition are missing, these are not marked in the Available in Memory? column.
If just an index is missing however, the columnd in the Available in Memory? column are marked, as indexes are not listed separately.
If you click on the entry of the unloads for a given object, the system displays details for unloads of this object.
You can select one or more tables. As an administrator, you can select additional functions by choosing . These are normally not necessary. You can select the following functions:
Load Tables into Memory
Unload Tables from Memory
Activate Early Unload: This setting overwrites the unload flags that are pre-defined for each object type. If a memory bottlneck occurs after you have made this setting, the selected tables are suppressed in accordance with the LRU algorithm (Least Recently Used)more suppressed than other tables with the same classification (Early Unload Activev or Early Unload Not Activev).
Deactivate Early Unload