Exploration Tool for Collaborative Provisioning and DVFS in Heterogeneous Cloud Environments

• Lucas Rister Machado UFSM
• Michael Guilherme Jordan UFRGS
• Antonio Carlos Schneider Beck UFRGS
• Mateus Beck Rutzig UFSM

Resumo

Cloud Warehouses are evolving with diverse computational resources, including CPUs, GPUs, and accelerators, catering to a multitude of tenant applications. While this hetero-geneity promises improved performance and energy efficiency, harnessing its full potential poses challenges due to dynamic workload characteristics and variable application demands. To address this, scheduling approaches combined with optimization techniques like Dynamic Voltage and Frequency Scaling (DVFS) are crucial. However, integrating these approaches effectively can be complex, potentially leading to conflicts and diminished benefits. This paper proposes an Energy Aware Collaborative Provisioning Framework target for heterogeneous CPU-GPU cloud nodes which offline tailors the best provisioning strategy considering the target hardware to optimize the makespan and dynamically tunes the voltage and frequency scaling to reduce energy consumption. Results show the importance of strategically aligning DVFS policies with provisioning algorithms, showcasing up to a 16.08 x improvement in Energy-Delay Product (EDP), underscoring significant gains in energy efficiency and performance.

Palavras-chave: Runtime, Scheduling algorithms, Heuristic algorithms, Collaboration, Voltage, Dynamic scheduling, Systems engineering and theory, Energy efficiency, Hardware, Optimization