Using the peer-to-peer (DSRM) model for distributed resource recovery, an application's work request across distributed systems is connected by Systems Network Architecture (SNA) LU 6.2 conversations handled by a communications manager such as APPC/MVS. Each work request has a logical unit of work identifier (LUWID).

When the first protected conversation is allocated for a work request, RRS generates an LUWID for it. The resource manager handling the outbound protected conversation uses a call to the Retrieve_Work_Identifier service to obtain the LUWID and passes the LUWID in the conversation. When the application accepts an inbound protected conversation, the resource manager handling it uses a call to the Set_Work_Identifier service to set the LUWID.

A resource manager can also use a call to the Set_Work_Identifier service to set the next LUWID for transaction chaining. In this case, RRS generates the LUWID for the next UR based on the current and next LUWIDs for the current UR. RRS generates the LUWID before returning control to the application program following a commit or backout call and before invoking any COMPLETION exit routines.

Resource managers in distributed processing need the following services:

For more information, see Callable resource recovery services.

A resource manager that has taken the server distributed syncpoint resource manager (SDSRM) role for an interest in a UR additionally needs the following calls:

For information on the calls, see Callable resource recovery services.