Creating Repositories from Multidimensional Data Sources Schedule: Timing Topic 25 minutes Lecture 10 minutes Practice 35 minutes Total

Oracle BI Suite EE 10g R3: Build Repositories 18 - 2 Objective After completing this lesson, you should be able to create a repository using a multidimensional data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 2

Oracle BI Suite EE 10g R3: Build Repositories 18 - 3 Overview You can use the Administration Tool to add a multidimensional data source to the Physical layer of a repository. The only currently available data sources that are compliant with XML for Analysis (XMLA) are: Essbase Microsoft Analysis Services SAP/Business Warehouse (SAP/BW) Data from multidimensional sources can be displayed on an Oracle Business Intelligence (BI) interactive dashboard. Overview You use the Administration Tool to add a multidimensional data source to the Physical layer of a repository. The ability to use multidimensional data sources allows the Oracle BI Server to connect to sources such as Essbase, Microsoft Analysis Services, and SAP/Business Warehouse (SAP/BW) to extract data. You can use data from these sources to build requests in Oracle BI Answers, which can be displayed on an Oracle BI interactive dashboard. Note that Oracle Business Intelligence Suite Enterprise Edition 10g, Release 3 (BI EE 10g) delivers significant product enhancements to further enable enterprise-wide BI, including integration with Oracle online analytical processing (OLAP). In this release, Oracle’s native multidimensional data model—the analytic workspace (AW)—is made accessible to BI EE 10g by creating the required metadata in Oracle BI Administration Tool. The AW data is exposed to the BI EE 10g product stack and the OLAP engine is leveraged for analysis of that data. Oracle OLAP and AW are not covered in this course. Instead, refer to the Oracle by Example (OBE) titled Using Oracle OLAP With Oracle BI Enterprise Edition 10g Release 3 at the following site: http://www.oracle.com/technology/obe/obe_bi/bi_ee_1013/olap/index.html Oracle BI Suite EE 10g R3: Build Repositories 18 - 3

XML for Analysis (XMLA) Oracle BI Server connects to the multidimensional source using the XMLA-standard protocol: This requires that a fully functional Web services interface be available with the target data source. The standard dictates the various protocols that the Oracle BI Server can use to connect to the target and query data. XML for Analysis The Oracle BI Server connects to the multidimensional source using the XMLA-standard protocol. This requires that a fully functional Web services interface is available with the target data source. The standard dictates the various protocols that the Oracle BI Server can use to connect to the target and query data. Importing data from a multidimensional source creates the metadata of the data source in the Oracle BI repository. XMLA is a joint Hyperion and Microsoft specification for an open API that is designed to standardize the data access interaction between a client application and an analytical data provider working on the Web. Oracle BI Suite EE 10g R3: Build Repositories 18 - 4

Multidimensional Versus Relational Data Sources The primary differences between setting up multidimensional data sources and setting up relational data sources are in the Physical layer. The setup processes in the business model and presentation layers for multidimensional data sources and relational data sources are almost identical. Multidimensional Versus Relational Data Sources The primary differences between setting up multidimensional data sources and setting up relational data sources are in the Physical layer. The setup processes in the business model and presentation layers for multidimensional data sources and relational data sources are almost identical. Oracle BI Suite EE 10g R3: Build Repositories 18 - 5

Overview: Importing Multidimensional Data Sources During the import process, each cube in a multidimensional data source is created as a single physical cube table. Oracle BI Server imports the cube, including its metrics, dimensions, and hierarchies. After importing cubes, make sure that: The physical cube columns have the correct aggregation rule The default member type ALL is correctly imported for a hierarchy Overview: Importing Multidimensional Data Sources The Oracle BI Server uses XMLA standards to import data from a multidimensional data source to the Oracle BI repository. During the import process, each cube in a multidimensional data source is created as a single physical cube table. The Oracle BI Server imports the cube, including its metrics, dimensions, and hierarchies. After importing the cubes, you need to make sure that the physical cube columns have the correct aggregation rule and that the default member type ALL is correctly imported for a hierarchy. Oracle BI Suite EE 10g R3: Build Repositories 18 - 6

Considerations: Importing Multidimensional Data Sources Oracle BI Server imports only those dimensions and hierarchies that are supported by Oracle BI. If a cube has a ragged hierarchy or a parent-child hierarchy, it is not imported. Measure hierarchies are not imported or supported. It is recommended that you remove hierarchies and columns from the Physical layer if they will not be used in the business model. This eliminates maintaining unnecessary objects in the Administration Tool. It may result in better performance Considerations: Importing Multidimensional Data Sources Some companies model business hierarchies in relational databases using a table structure in which each row contains the key of its parent. Because different branches of such a hierarchy may have different depths from root to leaf, they are sometimes called “ragged hierarchies.” While relational databases can model ragged hierarchies very easily with the recursive join on the parent organization key, it is difficult using standard SQL to traverse and query such a hierarchy. Therefore, ragged hierarchies pose a problem for virtually all SQL-generating BI platforms. The Oracle BI Server only imports the dimensions and hierarchies that are supported by Oracle BI. Therefore, if a cube has a ragged hierarchy or a parent-child hierarchy, it is not imported. Additionally, measure hierarchies are not imported or supported. It is recommended that you remove hierarchies and columns from the physical layer if they will not be used in the business model. This eliminates maintaining unnecessary objects in the Administration Tool and may result in better performance. Oracle BI Suite EE 10g R3: Build Repositories 18 - 7

Oracle BI Suite EE 10g R3: Build Repositories 18 - 8 ABC Example Create a business model using an Essbase multidimensional data source. ABC Example In the example, you create a business model in an Oracle BI repository using an Essbase multidimensional data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 8

Creating a Multidimensional Business Model Import a physical schema. Set up the connection pool. Verify the import. Verify imported hierarchies and levels. Verify imported measures. Work with preaggregated measures. Update member counts. View members. Add a hierarchy. Create the Business Model and Mapping (BMM) layer. Create the Presentation layer. Verify the results. Creating a Multidimensional Business Model This slide lists the steps for creating a business model using a multidimensional data source. Each step is presented in detail in the slides that follow. Oracle BI Suite EE 10g R3: Build Repositories 18 - 9

1. Import a Physical Schema Select File > Import from XMLA. Select the catalogs or cubes to import. XMLA provider URL Data source Username and password 1. Import a Physical Schema a. From your data source administrator, get the URL connection string, username, and password for the data source. b. In the Administration Tool, select File > Import from XMLA. c. In the Import From XMLA dialog box, complete the fields: - Select the XMLA provider. - Enter the URL for the Web service that acts as the XMLA provider. - Use the Update button to populate the Data Source field. - Enter the username and password. d. Select the catalogs (databases) or cubes to import and click Import. e. Enter a name in the Target Database field or browse to use an existing database object in the Physical layer. Oracle BI Suite EE 10g R3: Build Repositories 18 - 10

2. Set Up the Connection Pool Some connection pool properties are unique to multidimensional data sources. XMLA call interface XMLA virtual directory of the machine hosting the cube Vendor-specific information used to connect to the multidimensional data source 2. Set Up the Connection Pool Some connection pool properties are unique to multidimensional data sources: Call interface: XMLA URL: The URL to connect to the XMLA provider. It points to the XMLA virtual directory of the machine hosting the cube. Data Source: The vendor-specific information used to connect to the multidimensional data source. Consult your multidimensional data source administrator for setup instructions because specifications may change. Catalog: The list of catalogs available, if you imported data from your data source. The cube tables correspond to the catalog you use in the connection pool. Catalog Oracle BI Suite EE 10g R3: Build Repositories 18 - 11

Oracle BI Suite EE 10g R3: Build Repositories 18 - 12 3. Verify the Import When you import the physical schema, Oracle BI Server imports the cubes, including metrics, hierarchies, and levels. Database object Catalog Cube table Balanced hierarchies 3. Verify the Import When you import the physical schema, Oracle BI Server imports the cube, including its metrics, hierarchies, and levels. Each multidimensional catalog in the database can contain multiple physical cubes. You can import one or more of these cubes into your BI repository. You can create a cube table manually. However, it is recommended that you import cube tables and their components. Each cube from a multidimensional data source is set up as a physical cube table, a type of physical table. It has all the capabilities of a table such as physical cube columns, keys (optional), and foreign keys (optional). It also has cube-specific metadata such as hierarchies and levels. In the Physical layer, a physical cube table looks like a regular table but has a different icon. Columns also have unique cube icons: Key icons represent attributes that are part of the hierarchy. Columns with cube icons represent attributes that are not part of the hierarchy. Columns with cube icons plus the sigma sign represent either additive measures or calculated members. Measure cube columns Oracle BI Suite EE 10g R3: Build Repositories 18 - 12

4. Verify Imported Hierarchies and Levels In the Physical Cube Table dialog box, click the Hierarchies tab to verify that the import process imported the levels correctly. Hierarchies Levels 4. Verify Imported Hierarchies and Levels In the Physical Cube Table dialog box, the Hierarchies tab lists the dimensional hierarchies in the cube. In this dialog box you can add, edit, or remove hierarchies (buttons for these actions are not shown in the screenshot). To verify a hierarchy, select it and click Edit, or double-click the hierarchy. In the Hierarchy dialog box, verify that the levels are correct. The Hierarchy dialog box lists all the defined levels for the selected hierarchy. The highest level in the hierarchy should be the first (highest) item in the list. If you need to reorder the hierarchy levels, select a level and click Up or Down to correct the order of the levels. There must be multiple levels and you must select a level for the buttons to be available. You can also reorder, add, edit, or remove levels. Note the multidimensional level icon in the screenshot. This confirms that these columns have been identified as part of the hierarchy. The “Default Member type ALL” check box should always be selected by default. This is for performance reasons. This check box helps Oracle BI Server rewrite more efficient Multidimensional Expressions (MDX) when sending logical queries. If you delete property or key columns from a level, the association is deleted and the column changes to a measure under the parent cube table. Ragged hierarchies and unbalanced hierarchies are not yet supported in Oracle BI. These types of hierarchies are currently ignored by the import process. Reorder, add, edit, or remove levels. Oracle BI Suite EE 10g R3: Build Repositories 18 - 13

5. Verify Imported Measures Verify that the aggregation rule for a physical cube column is set correctly. 5. Verify Imported Measures Use the following guidelines to verify and assign the aggregation rule correctly: Verify aggregation rules after importing a cube. Typically, aggregation rules are assigned correctly when you import the cube. However, if a measure is a calculated measure, the aggregation rule is reported as None. Therefore, you should examine the aggregation rule for all measures after importing a cube to verify that the aggregation rule has been assigned correctly. For all measures assigned an aggregation rule value of None, contact the multidimensional data source administrator to verify that the value of the aggregation rule is accurate. If you need to change the aggregation rule, you can change it in the Physical Cube Column dialog box. If you build the measures manually, set the aggregation rule to match its definition in the multidimensional data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 14

6. Work with Preaggregated Measures In a multidimensional data source, some cubes contain very complex, multilevel-based measures. If the Oracle BI Administrator assigns an aggregation rule of Aggr_External, BI Server bypasses its internal aggregation mechanisms and uses the preaggregated measures. 6. Work with Preaggregated Measures In a multidimensional data source, some cubes contain very complex, multilevel-based measures. If the Oracle BI Administrator assigns an aggregation rule of Aggr_External, the BI Server bypasses its internal aggregation mechanisms and uses the preaggregated measures. When imported, these measures are assigned an aggregate value of None. The following are some guidelines for working with preaggregated measures: External aggregation applies to only multidimensional data sources that support these complex calculations. You cannot assign external aggregation to measures from standard relational data sources. If the measure is supported and can be mapped to a relational database, then it is not complex and does not require external aggregation. You cannot mix noncomplex measures from standard data sources (relational) with complex measures from multidimensional data sources. You can mix noncomplex measures from standard data sources (relational) with noncomplex measures from multidimensional data sources if they are aggregated through the Oracle BI Server. Oracle BI Suite EE 10g R3: Build Repositories 18 - 15

Oracle BI Suite EE 10g R3: Build Repositories 18 - 16 7. Update Member Counts To update member counts, the repository must be open in online mode. …and select Update Member Count. Right-click one or more physical objects… 7. Update Member Counts You must open the repository in online mode to update member counts. To determine if counts need to be updated, place your cursor over the hierarchy or level name. A message informs you that the counts need to be updated or showing you when they were last updated. When you update member counts, the current number of members are returned from the selected hierarchy. After the member count is updated successfully, it appears in a message when you place the cursor over the hierarchy or level name. The message appears in the following syntax: <hierarchy name> (<x> members, last updated <time stamp>) Place cursor here to see member count. Oracle BI Suite EE 10g R3: Build Repositories 18 - 16

Oracle BI Suite EE 10g R3: Build Repositories 18 - 17 8. View Members To view members, the repository must be opened in online mode. 1. Right-click a hierarchy or level. 2. Select View Members. 8. View Members To view members, the repository must be opened in online mode. You can view members of hierarchies or levels in the physical layer of repositories. The list of members by level in the hierarchy can help you determine if the XMLA connection on the server is set up properly. You may want to reduce the time it takes to return data or the size of the returned data by specifying a starting point (Starting from option) and the number of rows you want returned (Show option). Click Query and open the NQQuery.log file to view query results. 3. View members from the hierarchy. Oracle BI Suite EE 10g R3: Build Repositories 18 - 17

Oracle BI Suite EE 10g R3: Build Repositories 18 - 18 9. Add a Hierarchy Create a hierarchy object. Add levels and columns. Modify the hierarchy. 9. Add a Hierarchy This slide lists the steps for adding a hierarchy in the Physical layer. Each step is presented in detail in the slides that follow. Oracle BI Suite EE 10g R3: Build Repositories 18 - 18

9a. Create a Hierarchy Object Right-click a cube table object and select New Object > Hierarchy. 9a. Create a Hierarchy Object Most hierarchies are imported to the physical layer. Columns associated with a hierarchy that is not imported are not imported. If users need access to columns that are not imported, they should first add these columns to the Physical layer and then associate them with a level in a hierarchy. To create a hierarchy, right-click a cube table object, select New Object > Hierarchy, and enter values in the Hierarchy dialog box. Click Add to add levels to the hierarchy. Click Add to add levels to the hierarchy. Oracle BI Suite EE 10g R3: Build Repositories 18 - 19

9b. Add Levels and Columns To add a column to a level, in the Physical Level dialog box, click Add. Then, in the Browse dialog box, select the column and click Select, or double-click the column to select it. Each level in a hierarchy has a level key. The first cube column associated with (added to) the level of a hierarchy is the level key. This must match with the data source definition of the cube. The data source cube table cannot set one column as a level key and the Oracle BI physical layer table set a different column as a level key. The icon for the column that you select first changes to the key icon after it is associated with the level of a hierarchy. When you select columns to add to a hierarchy, it is recommended that you select them in hierarchical order, starting with the highest level. If you select multiple columns and bring them into the hierarchy at the same time, the order of the selected group of columns remains the same. After adding columns to the hierarchy, you can change the order of the columns in the Browse dialog box. Oracle BI Suite EE 10g R3: Build Repositories 18 - 20

Oracle BI Suite EE 10g R3: Build Repositories 18 - 21 9c. Modify the Hierarchy After a hierarchy is created, you can reorder, add, edit, or remove levels. 9c. Modify the Hierarchy After a hierarchy is created, you can reorder, add, edit, or remove levels. If a query does not explicitly refer to a member of a hierarchy, a default member must be used. Therefore, every hierarchy must be associated with a default member, typically the ALL member. The Hierarchy dialog box contains a check box (Default member type ALL) that you should select when you want to designate the ALL member as the default. Oracle BI Suite EE 10g R3: Build Repositories 18 - 21

10. Create the Business Model and Mapping Layer Setting up the Business Model and Mapping layer for multidimensional data sources is similar to setting up the logical layer for a relational data source. Drag the cube to the BMM layer. 10. Create the Business Model and Mapping Layer Setting up the Business Model and Mapping (logical) layer for multidimensional data sources is similar to setting up the logical layer for a relational data source. To create the business model layer, you can drag the physical layer cube to the logical layer. When you drag from the Physical layer, logical tables, dimensions, and relationships are created automatically. You can then modify objects in the Business Model and Mapping layer just as you would with a relational data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 22

11. Create the Presentation Layer Setting up the Presentation layer for multidimensional data sources is similar to setting up the Presentation layer for a relational data source. Drag the business model to the Presentation layer. 11. Create the Presentation Layer Setting up the Presentation layer for multidimensional data sources is similar to setting up the Presentation layer for a relational data source. To create the Presentation layer, you can drag the business model to the Presentation layer. You can then modify objects in the Presentation layer just as you would with a relational data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 23

Oracle BI Suite EE 10g R3: Build Repositories 18 - 24 12. Verify the Results Build and execute the query in Oracle BI Answers. Verify SQL in NQQuery.log. 12. Verify the Results To verify the results, build and execute queries in Oracle BI Answers and then verify the SQL in the query log. Oracle BI Suite EE 10g R3: Build Repositories 18 - 24

Oracle BI Suite EE 10g R3: Build Repositories 18 - 25 Summary In this lesson, you should have learned how to create a repository using a multidimensional data source. Oracle BI Suite EE 10g R3: Build Repositories 18 - 25

Oracle BI Suite EE 10g R3: Build Repositories 18 - 26 Practice 18-1: Creating a Repository Using a Multidimensional Data Source This practice covers the following topics: Importing an Essbase cube to the Physical layer Building a business model Verifying the results Oracle BI Suite EE 10g R3: Build Repositories 18 - 26