Routing objects on action paths in a distributed computing system

TL;DR: In this paper, the authors describe a system for routing an object along an action path (itself an "object") which defines the logical path to be traversed by the object.

Abstract: A system for routing an "object" (in the sense that object means an abstraction that encapsulates data in a known way, with a known list of operations or methods to access the data, and the object has a unique identity, is mobile, and possibly persistent). The "object" is routed in a distributed computing system along an action path (itself an "object") which defines the logical path to be traversed by the object. The action path consists of action stops naming or describing functionally principals (people or automated mechanisms) required to act upon the routed object in a prescribed order. The object routing system propagates the object along this action path, and monitors and controls its progress until it completes the path. The system includes mechanisms of dispatching the routed object between principals, finding the principals required to act on the routed object, notifying the principals in turn of their required action, and potentially relocating the routed object to the nodes of the principals. Optionally, the object routing system may use mechanisms for nagging principals about pending actions (if no progress occurs within a specified period), reporting such lack of action or progress to other principals, supporting the sharing of an action path by multiple routed objects, and facilitating the routing of an object in parallel to multiple principals. This object routing system is constructed as a generic service layer above services for object management, migration, persistence and interobject communication.