(DI-2208) BW Data Management

Table of Contents:

Busy and quiet times of queries and DTPs

The analysis provides an overview of busy and quiet times of queries and DTP loads which were executed on all InfoProviders on a monitored system. The Heatmap shows a total number of queries, DTPs, or their sum. You can switch it with buttons Combined (Query & DTP), DTP, or Query. Data is aggregated by weekdays and hours and it can be filtered by specific InfoProvider or by analyzed period. This analysis can help you to find quiet hours e.g., for operations that need more system resources. Also, the busiest hours can be examined and their content can be split into other less busy hours.

Busy and quiet times of queries and DTP analysis chart


InfoProviders size

The analysis provides an overview of all InfoProviders in the system divided into 3 categories. Each category represents the size of InfoProviders that correspond to a particular group (e.g. "10%" group represents the first 10% largest InfoProviders and their corresponding size). By clicking on a bar, the application will be redirected to the "Top InfoProviders" analysis, where the "Top %" filter will be applied accordingly.

.

InfoProviders size analysis chart


You can switch to the table view for more detailed information where data can be downloaded into a CSV file for further processing.

InfoProviders by year

This analysis provides an overview of InfoProviders in the system. Two charts are displayed here:

  • Distribution by InfoProvider type - displays the whole system per InfoProvider type
  • InfoProviders - objects in the chart are sorted by their size by default and can be further filtered

Both charts always display the last 4 years + all others merged into other groups (e.g., 2013 and lower) based on the amount of InfoProvider data that belong to the group.

.

InfoProviders by year analysis chart 


You can filter data and also switch to a table view for more detailed information where data can be downloaded into a CSV file for further processing.

It is possible to switch the color mode of the bottom chart to "Query usage". This colorizing mode groups data in InfoProvider by their respecting reporting usage. All data that were reported on less than 10 times are colorized with blue color and all data reported on more than 1000 times are colorized with red color. Warm data colorized with orange color belongs to the category between these two threshold values. 

.

InfoProviders by query usage analysis chart


You can change these threshold values by clicking on the  button.


NLS potential

Based on the analysis performed in the system, the visualization provides you with an overview of the potentials for archiving particular object types (such as InfoCubes, DataStore objects, Write-Optimized DSOs) in form of column charts. 

.

NLS potential analysis chart

You have the possibility to change archiving settings. The analysis recalculates the archiving potential accordingly by the button. 

NLS potential analysis settings 


Table view provides a detailed archiving potential on the object level. You can download data into a CSV file for further processing. 

NLS potential analysis table view 


Data distribution analyses

These analyses provide information about how data is distributed in InfoProviders based on a specified characteristic and based on how many times these data areas were accessed by reporting (queries). The number of Data distribution analysis tiles depends on the settings on the central system. By default, the BW analysis is delivered with only one predefined Data distribution analysis. The Data distribution analysis tile displays the number of analyzed InfoProviders that contain the specified characteristic. The figure below displays the temperature-based data categorization which is a type of data distribution and is by default included in the package.

Data Distribution tile

By clicking on a tile you can view detailed results of the specified Data distribution analysis. 

The first load of the analysis results can last multiple seconds due to the post-processing of the results from the system.

The result table contains multiple columns. Below are columns that are present in all Data distribution analysis results:

ColumnDescription
ObjectContains the technical name of an InfoProvider
TypeDisplays a type of an InfoProvider (InfoCube, DSO, aDSO, WODS)
DescriptionLong description of an InfoProvider
Size [MB]

Total object size in MB

  • InfoCube: F table + E table + dimension tables
  • DSO / WODS / ADSO: new table + active table

Table size calculation differs by various database types

  • HANA database: main memory + delta memory + index
  • Other databases: table size (without index)
RowsNumber of rows in InfoProvider data tables (same tables are used as in the size calculation)
Number of queries

The total number of queries that read data in an InfoProvider

  • This column can contain a value > 0, also in the case, the InfoProvider is empty. This event occurs when a query uses a MultiProvider which includes this InfoProvider
Trend - Number of queriesVisualization displaying the usage of data areas - values are rounded to natural numbers
Trend - Size [MB]Visualization displaying the size of data areas - values are rounded to natural numbers
Split characteristic

Describes applied time characteristic

  • if the characteristic is empty, then the Data distribution analysis doesn't identify a suitable characteristic, and the distribution of the size and number of queries isn't successful
Archived size [MB]

Size of unpacked archived data in MB 

  • size of data in an online database that would be occupied, if all archived data would be reloaded
DTP source countHow many times an InfoProvider is used as a source in DTP load requests
Last DTP exec.The timestamp of the last DTP load request execution
DTP dest. countNumber of DTP requests loaded into an InfoProvider (including reloaded requests from an archive)
DTP dest. relv

Number of requests (DTPs for InfoCubes, activation requests for DSOs) relevant for straggler analysis

  • by a DSO object - request is relevant when a Changelog contains records of this request
  • by an InfoCube object - request is relevant when request records are not compressed
  • by a WODS object - requests aren't relevant, as WODS can be archived only based on request creation date
  • by an ADSO object - based on the type of the object, the ADSO behaves the same as the equivalent object type (DSO, WODS, InfoCube)
    • as ADSO object can have various settings. In general, the analysis collects requests for the following object types
      • Data warehouse - data mart
        • information is gathered from the inbound table
      • Corporate memory – reporting capabilities
        • information is gathered from the inbound table
      • Data warehouse – delta calculation
        • information is gathered from the Changelog table
DTP stragglers

Number of requests with straggler records (see more details)

  • subset of requests relevant for straggler analysis, which contain at least one straggler record
DTP stragg. rowsNumber of rows that represent straggler records
Number of Lookups

Number of Lookups that are present in the system for an InfoProvider

  • number of Lookups in the source code


Data distribution analysis standard view


The columns number of queries and size of data areas depend on data distribution settings of a particular analysis. These columns are created based on data groups, you specify in your analysis. See the chapter (DI-2208) Define the Data Distribution Analysis for details about how to define data distribution and its respective data groups for analysis. Each of these columns contains a sum (for sizes) and sum (for the use) of the corresponding data area values. The footer of the table displays the number of records that were already loaded to the table. This number will gradually increase, as you scroll down in the table.

Note that data groups will be empty, if the distribution analysis could not identify the split characteristic for a particular InfoProvider and distribution wasn't performed. For more information on why the split characteristic was not identified, see the chapter (DI-2208) Define the Data Distribution Analysis.

You can filter the content of a table column by clicking on the column header. There is also a possibility to download the table content (without chart columns) into a CSV. file by clicking on  toolbar button. You can also hide/show specific column groups by clicking on column group in View options toolbar. You can sort results based on the field of your preference, by clicking again on the column header:

Sorting and filtering directly according to the column header

You can click on any InfoProvider line to display queries that were executed on the InfoProvider. Each line in this detail view contains a query that was executed with a filter selection that was used during its execution (only filters for the same characteristic as was used for data distribution are displayed). 

InfoProvider query execution details