The Conceptualization of a Configurable Multi-party Multi-message Request-Reply Conversation

Abstract

Abstract. Organizations, to function effectively and expand their boundaries, require a deep insight into both process orchestration and choreography of cross-organization business processes. The set of requirements for service interactions is significant, and has not yet been sufficiently refined. Service Interaction Patterns studies by Barros et al. demonstrate this point. However, they overlook some important aspects of service interaction of bilateral and multilateral nature. Furthermore, the definition of these patterns are not precise due to the absence of a formal semantics.In this paper, we analyze and present a set of patterns formed around the subset of patterns documented by Barros et al. concerned with RequestReply interactions, and extend these ideas to cover multiple parties and multiple messages. We concentrate on the interaction between multiple parties, and analyze issues of a non-guaranteed response and different aspects of message handling. We propose one configurable, formally defined, conceptual model to describe and analyze options and variants of request-reply patterns. Furthermore, we propose a graphical notation to depict every pattern variant, and formalize the semantics by means of Coloured Petri Nets. In addition, we apply this pattern family to evaluate WS-BPEL v2.0 and check how selected pattern variants can be operationalized in Oracle BPEL PM.

Interacting services: From specification to execution

“…In addition to the Service Interaction Patterns framework [7] that is used in this paper, there is a family of interaction patterns reported by Mulyar et al [42]. Here, different variation points for multi-party multi-message conversations are identified, also allowing for a detailed analysis of choreography languages.…”

CorPN: managing instance correspondence in collaborative business processes

“…Mulyar, Aldred, and van der Aalst investigated the message multicasting patterns in service interactions [19]. Their research took into account the factors of message queuing, sorting and indexing, possibilities of non-responding parties and missing replies, etc.…”