Electricity Cost Minimization for Multi-Workflow Allocation in Geo-Distributed Data Centers

TL;DR

This work tackles electricity cost minimization for multi-workflow allocation in geo-distributed data centers under dynamic price conditions and deadline constraints. It introduces ECMWS, a four-stage framework integrating Contention-aware Workflow Sequencing, BLDP-based sub-deadline partitioning, task ranking-based sequencing, and a graph-embedding+PPO-based resource allocator with a backup DARA strategy. Key contributions include dual graph embeddings for tasks and resources, an Actor-Critic (RAPPO) policy for cost-aware allocations, and extensive parameter calibration plus ablation studies showing the importance of each module. Empirical results on cross-domain DCs demonstrate that ECMWS achieves meaningful cost savings (over 15%) while maintaining practical computation times, indicating strong potential for real-world, price-aware, geo-distributed workflow scheduling.

Abstract

Worldwide, Geo-distributed Data Centers (GDCs) provide computing and storage services for massive workflow applications, resulting in high electricity costs that vary depending on geographical locations and time. How to reduce electricity costs while satisfying the deadline constraints of workflow applications is important in GDCs, which is determined by the execution time of servers, power, and electricity price. Determining the completion time of workflows with different server frequencies can be challenging, especially in scenarios with heterogeneous computing resources in GDCs. Moreover, the electricity price is also different in geographical locations and may change dynamically. To address these challenges, we develop a geo-distributed system architecture and propose an Electricity Cost aware Multiple Workflows Scheduling algorithm (ECMWS) for servers of GDCs with fixed frequency and power. ECMWS comprises four stages, namely workflow sequencing, deadline partitioning, task sequencing, and resource allocation where two graph embedding models and a policy network are constructed to solve the Markov Decision Process (MDP). After statistically calibrating parameters and algorithm components over a comprehensive set of workflow instances, the proposed algorithms are compared with the state-of-the-art methods over two types of workflow instances. The experimental results demonstrate that our proposed algorithm significantly outperforms other algorithms, achieving an improvement of over 15\% while maintaining an acceptable computational time. The source codes are available at https://gitee.com/public-artifacts/ecmws-experiments.

Figures (7)

• Figure 1: System Architecture.
• Figure 2: Task diagram example.
• Figure 3: Resource diagram example.
• Figure 4: Actor-Critic network architecture.
• Figure 5: Electricity price fluctuations in the four DCs.