Methodologies for branching

Before you branch a project, you must decide how you want to share the metadata in your project. This section describes some common methodologies for sharing metadata.

Hub and spoke

In hub and spoke, a project uses common metadata that must be shared by all functional areas. The root project consists of a fully modeled physical layer containing the objects that all functional areas require. The root project is branched for each functional area. Each functional area can create its own branches if there are multiple people working on it. At any time, a functional area modeler can merge a branch back into the root project to update the root project, and then branch again to receive updates. Objects that are common to all functional areas are kept in the root project.

Functional area specific metadata

In functional area specific metadata, there is little or no common metadata in the project. Each functional area develops their own objects in the project independently. Each functional area is unaware of the objects in the other functional areas. The master modeler controls merging of the branches to prevent each functional area from seeing objects in the other functional areas.

Distribution by layers

In distribution by layers, the metadata is organized in layers. Each layer requires access to the metadata of the layer above it. For example, a model contains three layers. The top layer is the root project, consisting of a fully modeled physical layer. The root project is branched to create the second layer, the intermediate branch. The intermediate branch contains a fully modeled development layer. The intermediate branch is branched to create the third layer, the presentation branch. The presentation branch contains a fully modeled business layer where reporting packages are defined and published.