ORCHESTRATION OF CUSTOMER INTERACTION WORKFLOWS IN ENTERPRISE APPLICATIONS

Abstract

Process automation is an important factor in the development of modern corporate systems, in particular e-commerce systems. One of the tasks that must be solved during development of such systems is the orchestration of customer interaction processes, which are asynchronous in nature. Modern popular orchestration frameworks (Airflow for example), are not very well adapted to simultaneously waiting for a response from the client for a large number of workflows. This leads to unnecessary costs of computing resources and a decrease in the economic efficiency of the system. This paper considers the task of building a system for orchestrating customer interaction workflows based on an event-driven approach. Instead of a sequential-like centralized execution of the graph model of the workflow, it is proposed to present the workflow as a sequence of events and the system's reactions to them, which eliminates the need to explicitly wait for a response from the client. In such model workflow operations will be performed during event processing, and the transition to the next operation will occur by sending a new message with a description of the operation. Centralized and distributed approaches for performing transitions between workflow operations are considered, the advantages and disadvantages of each of them are shown. The implementation of waiting for client response is also considered, for which it is proposed to store the message tied to the client session identifier in a specialized storage before starting interaction with client. And after receiving data from the client, add them to the message and send it back to the message queue to continue the workflow. Taking into account the possibility of using handlers written in different programming languages, JSON format based description of workflow model is proposed to be used. One of the approaches to building model description format is described and its use for demonstrating example of a workflow. The results of the study can be useful when creating systems for orchestrating workflows of interaction with the client.