Sustain-Cluster Documentation¶
Sustain-Cluster is a benchmark environment designed to evaluate dynamic workload distribution techniques for sustainable Data Center Clusters (DCC). It aims to reduce the environmental impact of cloud computing by distributing workloads within a DCC that spans multiple geographical locations. The benchmark environment supports the evaluation of various control algorithms, including reinforcement learning-based approaches.
This page contains the documentation for the GitHub repository and the paper “SustainCluster: Benchmarking Dynamic Workload Distribution Techniques for Sustainable Data Center Cluster” <https://openreview.net/forum?id=AgTASkryb6>_
Demo of Sustain-Cluster¶
A demo of Sustain-Cluster is given in the Google notebook below:
Key features of Sustain-Cluster¶
Dynamic geographic distribution of workloads across data centers in a cluster, driven by a central Top-Level Agent.
Dynamic temporal scheduling decisions by the Top-Level Agent, determining whether tasks should be executed immediately or deferred to optimize efficiency.
Incorporation of non-uniform computing resources, cooling capabilities, auxiliary power resources, and varying external weather and carbon intensity conditions.
A dynamic bandwidth cost model that accounts for the geographical characteristics and amount of data transferred.
Realistic workload execution delays to reflect changes in data center capacity and demand.
Support for benchmarking multiple heuristic and reinforcement learning-based approaches.
Customizability to address specific needs of cloud providers or enterprise data center clusters.
Sustain-Cluster provides a complex, interdependent, and realistic benchmarking environment that is well-suited for evaluating reinforcement learning algorithms applied to data center control. The ultimate goal is to optimize workload distribution to minimize the carbon footprint, energy usage, and energy cost, while considering various operational constraints and environmental factors.
Scheduling Strategy¶
Sustain-Cluster centralizes both geographic and temporal workload distribution decisions in the Top-Level Agent. At each scheduling step, the agent determines whether to execute a given task immediately or defer it, and selects the optimal data center for execution. This unified spatio-temporal approach ensures that all load-shifting logic is handled globally.