Show TOC

Analyzing Statistics Data for a MultiProvider QueryLocate this document in the navigation structure

You want to analyze the database access times for a query that you defined based on a MultiProvider. You want to know which of the relevant InfoProvider caused the longest database access times.

Caution

The analysis portrayed here is only suitable for smaller application scenarios. If the existing dialog processes are not sufficient to execute all the subqueries at the same time, the remaining subqueries are executed at a later time. In this case, the information below does not apply. Six dialog processes are available as standard.

The statistics level is set to 2 for the required query: The system records all data from the area for the front end and calculation layer, as well as data from the area for the aggregation layer and aggregation information.

To analyze the statistics data, you have the following options:

  • You can use the InfoCubes 0TCT_C01 (aggregated data) or 0TCT_C02 (detailed data) from the technical content.
  • You can display the detailed statistics data in the query monitor (with the debug option Display Statistics Data) on the tab page Aggregation Layer.
  • When you execute a Web application, you can display the statistics data by adding the parameter PROFILING='X' to the URL.
  • You can evaluate the underlying table RSDDSTAT_DM directly with the data browser (SE16).

Aggregated Statistical Data

000009000

Data Manager

2,801470

0

000009010

Total DBTRANS

0,000000

831

000009011

Total DBSEL

0,000000

104.264

Detailed Statistical Data

The detailed statistical data provides a view into the execution of MultiProvider queries:

  • On the basis of the MultiProvider, the system executes three subqueries in our example:
    • A query on the InfoProvider DataStore ODS1
    • A query on the InfoCube 0BWVC_006 (here the system uses the fact table F)
    • A query on the InfoCube IUSALES (BW Accelerator index '*$X' = BIA is used here)

    The access count 0 is the main process that the queries are split from.

  • The total ratio DBTRANS / DBSEL of 831 / 104.264 does not mean a lot for the MultiProvider query because it only totals the times of the relevant queries. In the case of the MultiProviders, the detailed statistical data offers more meaningful results. The ratio for the query on the InfoCube 0BWVC_006 is, for example, lower than 5 (75 / 300).
  • It is too complex to calculate the data manager total time using the individual times. If there are no more read access instances that were performed as dialog processes, the data manager total time is calculated using the longest data read time plus additional times (such as DM Preparation, times for some function calls, or for splitting subqueries). As a result, the sum of the detailed statistical times is always lower than the total time for the EVENTID 9000.