Skip to Main Content
Executing workflows on large-scale heterogeneous distributed computing systems is a challenging task. Albeit there have been many well elaborated approaches, they are often motivated by a certain class of applications and focus on some implementation specific problems. The work presented in this paper aimed at establishing a highly abstract coordination model for distributed workflow enactment where decentralized control, autonomy, adaptation to high dynamics and partial lack of information are of primary concerns. The model is based on a nature metaphor and envisioned as a chemical reaction where molecules react autonomously according to local and actual conditions. The execution model of such chemical workflow enactment is formalized using the γ-calculus. In the γ-calculus control, scheduling, dependencies, errors and the state of the computation are all represented in a single uniform declarative formalism that has a mathematically founded clear semantics. The paper shows that the abstract coordination model expressed in γ-calculus is able to grasp all aspects of such a chemical enactment, provides a more complex and adaptive framework than most current approaches, and actual realizations may be founded on it.