Answers are grounded in this paper and cite their sources.
This research addresses the workflow scheduling problem on shared computing units. A set of interdependent tasks, originating from a workflow or project, requires processing by computing units such as virtual machines, while respecting precedence constraints and capacity restrictions, for example, in terms of virtual central processing units. The computing units may be heterogeneous or homogeneous in their capacities and cost rates. This problem arises in cloud environments that allow multiple tasks to share resources to improve utilization and reduce energy consumption in accordance with a race-to-idle strategy. We present different mathematical model formulations for this integrated model of workflow scheduling and vector bin packing: a mixed-integer program based on the pulse disaggregated discrete-time formulation, a mixed-integer quadratically constrained program, and a constraint programming satisfiability model, which can be further extended to related scheduling settings. The formulations are evaluated under a makespan objective and under a weighted-sum objective combining makespan and active-unit energy cost. Computational experiments with the Gurobi Optimizer and the CP-SAT solver show that the pulse-based formulation provides the strongest bounds and proves optimality most often, whereas the constraint programming formulation quickly finds high-quality incumbent solutions for the combined objective.
No takes yet. Share an insight, caveat, or question.
Stieber et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: