The resources that a unit of recovery updates can be distributed; that is, they can reside on more than one system. A unit of recovery can be distributed among multiple resource managers. Those resource managers can be on a local system or even on another system in the network. Two programs that participate in distributed resource recovery are DB2® and IMS™.

For example, when multiple resource managers are involved, an application might do some DB2 work, and then some DFSMStvs work, and finally some IMS work. All of the work can be interrelated so that it is all under the umbrella of a single unit of recovery that RRS coordinates. The application completes its transactions and then calls RRS to commit or back out those transactions. In this way, the work within this unit of recovery spans several different resource managers. A unit of recovery can be distributed not only across multiple resource managers but across systems and networks.