The goal of managing data access in a distributed system is to maintain a consistent view of the data from all areas of the system. Different areas of the system need to make decisions based on what the data looks like to them. If different areas see the data differently, functionality will be affected. For example, if one subsystem in a core switching product sees a certain equipment as entered into the system database, but another subsystem does not, a command will be denied rendering that equipment useless.

A Resource Manager can be used to coordinate access to data and preserve data integrity. The Resource Manager needs to minimize data access concurrency by preventing access to data while events are pending. At the same time the Resource Manager needs to maximize throughput by allowing mulitple concurrent access to mutually exclusive data sets. The Resource Manager uses a system of mutexes to apply "locks" to data so that it can only be accessed by a single area at a time. Locks are applied on categories, or "classes", of provisioning events. A provisioning event is any system event, user driven or autonomous, which affects provisioning data. In the context of the Resource Manager, the arrival of a provisioning event into the system is called a "trigger".

The purpose of the Resource Manager is to apply locks to classes of triggers in such a way as to minimize concurrency and maximize throughput. The challenge of designing a Resource Manager is finding the correct balance between protecting data and allowing as many multiple concurrent processes as possible.