Cost-Optimal Cross-Cloud Data Transfer Scheduling in Jointcloud Environments with Time-Dependent Pricing
Year: 2026
Type: conference-paper
Abstract: JointCloud environments, including multi-cloud and federated cloud systems, increasingly rely on high-performance networks (HPNs) to support large-scale cross-cloud data transfers. In such settings, advance bandwidth reservation with timedependent pricing is essential for cost-efficient and predictable data movement, where the transfer cost depends on both dynamic link prices and path reconfiguration overhead. This paper investigates the optimal scheduling of the VPFB BRR-MinC, where VPFB (Variable Path, Fixed Bandwidth) allows routing paths to change across time slots while maintaining a constant reserved bandwidth, and BRR-MinC seeks a minimum-cost schedule for deadline-constrained data transfers. We formalize a timedependent cost model incorporating slot-varying edge weights and switching penalties, and prove that the problem is NPcomplete. To address the temporal coupling introduced by switching costs, we develop a segmentation-based dynamic programming framework and propose a scalable heuristic, Heu-VPFB-MinC-TD-S. Simulation results on an ESnet-inspired topology show that the proposed method achieves identical feasibility while reducing total transfer cost compared with a greedy baseline, at the expense of moderate additional runtime. These results demonstrate the effectiveness of segmentation-aware optimization for cost-efficient cross-cloud data transfer in JointCloud systems.
Authors: Liudong Zuo, Pan Lai, Michelle Zhu, Chase Wu
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204104016