To answer the other questions raised at the beginning of this chapter, from Administrator Workbench right-click the concerned InfoCube (IC_NEWBC2 in our example), and then select Manage. Screen 12.12 appears.
Screen 12.12 contains six areas:
We will discuss each of these topics in turn.
12.2.1 InfoCube Contents
The first tab in Screen 12.12 is Contents. Here we can display and delete the InfoCube's contents.
Step 1. To display the InfoCube contents, click .
Step 2. In Screen 12.13, we specify conditions regarding what contents to be displayed. Here are some options:
Step 3. To delete data from the InfoCube, click .
Step 4. In this pop-up window, we can specify the data to be deleted and how to delete them by clicking .
Step 5. Besides the conditions used to specify the data to be deleted, in this window, we can also do the following:
12.2.2 Indices and Statistics
The second tab in Screen 12.12 is related to performance. Here we can check, delete, and create indices and statistics.
Step 1. The status lights on the right should be green. If no index/ statistics data exist, the status lights will be red.
To delete indices from the fact table, click .
Step 2. Now the left status light turns red. To create the indices, click .
Step 3. Click to continue.
Step 4. Click to continue.
Step 5. Depending on the size of the InfoCube, it may take a while for the status light to turn green.
To improve data loading performance, it is a good practice to delete the indices, load the data, and then recreate the indices. To automate this job, click .
Rolling up aggregates will automatically repair the aggregates' indices.
Step 6. Check all options, and then click to save the settings and continue.
The next time when we load data into this InfoCube, we will receive a message like the one in Screen 12.23. To continue the data loading operation, click .
In the same way, we can check, delete, and create the BW statistics.
Step 7. If the DB statistics status light is red, click .
Step 8. Click to continue.
Step 9. Click to continue.
Step 10. Depending on the size of the InfoCube, it may take a while for the status light to turn green.
Each time when new data are loaded into the fact table, statistics should be refreshed. To automate this process, click .
Step 11. Check both options, and then click to save the settings and continue.
From now on, the DB statistics will be refreshed after each data load and each delta load.
12.2.3 Data Load Requests
The third tab in Screen 12.12, Requests, allows us to manage individual data load requests.
In BW, each data load from BW Scheduler counts as a request and is assigned a unique request number consisting of 30 characters. This request number is stored as an integer value called a request ID, in the data packet dimension of the InfoCube.
As an example, Screen 12.29 displays the contents of the data packet dimension table, /BIC/DIC_NEWBC2P. It shows the relationships among the request IDs (473, 474, and 479) and DIMIDs (1, 2, and 3).
The DIMIDs (41, 42, and 43) are used in the fact table /BIC/FIC_NEWBC2 for the request IDs (Screen 12.30).
As shown in Screen 12.31, if an InfoCube has aggregates, then a request is not available for reporting until it is rolled up into the aggregates.
To delete a request from the InfoCube, click . If the request has been rolled up into an aggregate or compressed, however, then you cannot delete it. If it has been rolled up, then the aggregate must be deactivated first. If it has been compressed, then all data in the fact table must be deleted first.
BW provides a useful function that allows us to check whether one data load request overlaps with another. The following is an example for request 474.
Step 1. Select request 474, and then click .
Step 2. A new session window opens and displays the request's 30-character request number,REQU_3I4VT73UXJDCQIO47YE4Q4QAI.
Click to check the overlapping.
The result contains information about the overlapping. If overlapping is present, the result also tells us whether we can delete the request (Screen 12.34).
12.2.4 Aggregate Rollup
The fourth tab, Rollup, in Screen 12.12 deals with the aggregate rollup. When we roll up data load requests, we roll them up into all the aggregates of the InfoCube.
Step 1. To roll up request 479, enter 479 into the Request ID and then click . BW will roll up request 479 and other requests below 479 into the aggregates.
Returning to the Requests tab and clicking to refresh the screen, notice that request 479 has been rolled up. Now it is available for reporting (Screen 12.36).
12.2.5 InfoCube Compression
The fifth tab in Screen 12.12, Collapse, focuses on the InfoCube compression.
Each InfoCube has two fact tables: the F fact table and the E fact table. We load data into the F fact table, where the data are grouped according to the individual load among requests. The F table allows us to check for overlapping requests and to delete unnecessary requests.
The F fact table degrades query performance, however. When executing a query, the OLAP processor must aggregate key figures to eliminate the request information in the data packet dimension.
To overcome this drawback, BW allows us to aggregate the F fact table and save the aggregated data in the E fact table in a process called the InfoCube compression. After the InfoCube compression, the OLAP processor will go to the E fact table to retrieve data. The procedure for performing this compression appears next.
Step 1. To compress request 479, enter 479 into the Request ID and then click . BW will compress request 479 and other requests below 479.
To delete records whose key figures are all 0, select the With zero elimination option.
After the compression, the F fact table /BIC/FIC_NEWBC2 contains no more data. Instead, the compressed data now appear in the E fact table, /BIC/EIC_NEWBC2. The E fact table has only 24 records of data, instead of 72 as shown in Screen 12.30. The data packet dimension column KEY_IC_NEWBC2P contains only zeros.
For performance and disk space reasons, it is recommended that you roll up a request as soon as possible and then compress the InfoCube.
To roll up or compress a request automatically when data are loaded into an InfoCube, select the Automatic request processing… item from the Environment menu on the InfoCube definition screen (see Screen 12.39).
In the pop-up window (Screen 12.40), check all options and click .
The order in which requests are rolled up and compressed is very important. After their compression into the E table, requests are no longer available in the F fact table and therefore cannot be rolled up into the aggregates. If we roll up requests after compressing them, the status under the Requests tab will indicate these requests are now in the aggregates – that is not true.
To delete a compressed request, we must either delete all of the E fact table data or use the request reverse posting function in Screen 4.8 by clicking .
12.2.6 InfoCube Reconstruction
The last tab in Screen 12.12, Reconstruct, allows us to reconstruct InfoCubes. Here we can reinsert the requests in PSA back into the InfoCube. The procedure follows.
Using this feature, we can also fill other InfoCubes with the requests, as long as the requests remain in PSA. This feature is particularly useful when the requests call for delta loads because we cannot request the same delta loads again from the source system.
Work InstructionsStep 1. Repeat the Screen 8.14 step to delete the IC_NEWBC2 data. BW removes all three requests (473, 474, and 479) from Screen 12.36.
Returning to the Requests tab and clicking to refresh the view, we see that the three requests have returned to the InfoCube. They are not rolled up and compressed, but are available for reporting.
Part I. Guided Tours
Business Scenario and SAP BW
Creating an InfoCube
Loading Data into the InfoCube
Checking Data Quality
Creating Queries and Workbooks
Managing User Authorization
Part II. Advanced Topics
Aggregates and Multi-Cubes
Operational Data Store (ODS)
Generic R/3 Data Extraction
Appendix A. BW Implementation Methodology
Appendix B. SAP Basis Overview
Appendix C. Glossary
Appendix D. Bibliography