Modeling Service Engagements in Dynamic Organizations: Multiagent Model and Architecture for Policy-Based Governance

Abstract

Service engagements arise commonly in business and scientific computing. A service engagement is characterized by autonomous parties coming together in a contractual arrangement to share resources or carry out tasks for one another. The autonomy of the participants is key, meaning that there is no unique locus for policy application. Yet, autonomy is not properly treated by current approaches for designing service engagements, which typically take the perspective of one of the participants.

We provide an agent-based conceptual model for specifying service engagements as arising within dynamic service organizations or Orgs. The atoms of a service engagement are formalized as commitments among the participants, to be created and manipulated as the engagement progresses. An Org scopes the commitments formed in an engagement. Orgs and their members are modeled as agents. A service contract provides a natural arms-length abstraction for modeling service engagements and is formalized as comprising a set of commitments among the contracting agents. An institution is a kind of an Org that acts as a social and legal context, within which Orgs arise modeling different service engagements.

We provide a policy-based architecture for the governance of service engagements. We propose innovations in policy-based architecture to model and govern complex Orgs. Traditional policy-based frameworks emphasize reactive behaviors wherein an external request causes a policy engine to compute a response. However, business service settings require richer policies and call for proactive behaviors. A business must not only respond to explicit requests, but also monitor its environment, collate events, and potentially act in anticipation of events in order to ensure that its policies are satisfied.

The core of this research is an approach to formalize service engagements based on commitments and to study their dynamics in the presence of policies specified by each of the participants. We describe a methodology for service engagement design using a set of design patterns that capture commonly occurring elements of service engagements. We have implemented MAVOS, a prototype of a policy-based multiagent system to model Orgs that demonstrates and evaluates our approach on realistic scenarios involving Orgs.