Many operations call for connecting to source data external to the cube. Connections and Datasources, which you create and save as reusable objects in Oracle Essbase, provide a way to do this efficiently.
If you have network connectivity between an external source of data and Essbase, you can define connections and Datasources in Essbase to easily "pull" data from the external source. If you have no network connectivity between Essbase and the external source of data, then you should stream data loads or dimension builds using the CLI tool, first creating a local connection, and then issuing the dataload or dimbuild command with the stream option.
A connection stores information about an external server and the login credentials that are required to access it. By defining one connection that can be used by multiple processes and artifacts, you can simplify many aspects of your analytics. For example, when it’s time to change a system password, you only need to update one connection.
A Datasource is another object that you can define once and reuse, to help you manage data flow into and out of your cubes. You can define a Datasource to represent any external source of data, whether a relational system, a table, a file, or another cube.
You can define one connection and use it to access multiple Datasources. For example, consider an external Oracle Database server that has separate tables for products, resellers, and sales territories. You need only one connection to access Oracle Database, but you might want to create unique Datasources to access each of the tables.
One use case in which you might define multiple Datasources per connection is as follows: if you use separate load rules to build each dimension in a cube, each rules file can be set up to access the relevant table in Oracle Database. For example, assume your cube has a Market dimension, and you regularly build dimensions using a Dim_Market load rule to populate the Market dimension from a SALES_TERRITORIES table. Likewise, you use a Dim_Product load rule to populate the Product dimension from a PRODUCT table. Both load rules can use the same connection, but because they draw from separate tables, you defined two different Datasources.
Historically, you needed to hard code connection and source data details into Essbase artifacts such as rule files, location aliases, and partitions. While hard coded information is still supported in these artifacts, you can work more efficiently if you define connections and Datasources globally (or, at the application level).