You use BW statistics to evaluate the fundamental functional areas of the Business Information Warehouse.
BW statistics provides you with the following options that allow you to evaluate data from both the OLAP processor and warehouse management. You are able to
· get an overview of how InfoProviders, InfoObjects, InfoSources, source systems, queries, and aggregates are used
· determine the system performance and improve it
· improve the way in which aggregates are selected and used and reduce the cost of updating them
BW statistics are delivered as part of technical content. The Technical Content role contains all the objects that are needed to implement BW statistics, in particular the BW statistics MultiProvider (previously called the BW statistics MultiCube).
You install the technical content in the same way that you install Business Content, in the Business Content functional area of the Administrator Workbench.
InfoPackages are not automatically collected together. Collect and transfer the InfoPackages by using the InfoPackage groups delivered.
The following diagram gives you an overview of the dataflow in BW statistics:
The data in BW statistics is saved and managed in the Business Information Warehouse.
When a query is executed, data is specified for the OLAP server and for access to the database. This data is saved temporarily once the navigation step has been completed. This is also the case when the ODBO (OLE DB for OLAP) interface is used. Additional data is collected when the aggregates are filled and rolled up after loading data into warehouse management.
It does not take long to calculate and save BW statistics data. However, the dataset can be considerable with larger installations. For this reason, the data input for each InfoProvider in each area of OLAP and warehouse management can be activated and deactivated individually. You are able to delete stored data. (See Activating Data Transfers for BW Statistics)
The individual InfoCubes are filled by the various InfoSources. This reduces the time needed to stage the data. Extractors are used to load the data.
The concept of the MultiProvider is used for reporting.
The following are delivered with technical content for use in BW statistics:
BW Statistics MultiProvider with InfoCubes
BW Statistics Queries with queries and diagrams
The types of questions that BW statistics helps you to answer can be grouped as follows:
· Which InfoCubes, InfoObjects, InfoSources, source systems, queries, and aggregates are currently being used in the system? How often are these objects being used? Which datasets are in use? Which users are currently working in the system?
· Are there any queries that have been running for longer than the fixed time for online processing allows? Are tasks that run for a long time, such as loading data, carried out when the system is not working to capacity? (See Optimization.)
· How does the data flow through the data warehouse? (Where does the data come from?) To which data targets is it moved?
· Which departments or users have used BW over a particular period of time (for example, during the last quarter, in the last year)?
· How has the workload for the database, for the OLAP processor, or for the frontend changed over time? What new demands can you expect for the future?
· From time to time a job that runs for a considerable amount of time has to be scheduled.
· Printing reports in the background.
· Loading large volumes of data.
· Rolling-up the data in aggregates.
When is the best time to schedule this kind of job? When is the system least busy? You can check the schedule later: Was this really the best time to schedule the job, or was there overlapping that lead to the system being overloaded?
· Which aggregates could you use to reduce the runtime of the queries? At what rate does the time taken to load the data, including the time it takes to roll-up the aggregates, increase?
· Which aggregates, InfoCubes, InfoObjects, or InfoSources are no longer used and can therefore be deleted? Are you able to or must you change the periodic loading of InfoCubes?
· Performance: To what extent do queries put pressure on the database, the OLAP server, and the frontend? Can this pressure be reduced by changing the definition of the queries?
· How much or how little is the InfoCube used, and how complicated is it to load data into the InfoCube?