Going Green for Less Green: Optimizing the Cost of Reducing Cloud Carbon Emissions

This artifact contains the GAIA scheduler's source code, the proposed scheduler in the paper: Going Green for Less Green: Optimizing the Cost of Reducing Cloud Carbon Emissions, and execution instructions. GAIA is written in Python and requires 1) pcluster: AWS ParallelCluster management command line interface and 2)PySlurm: Python-based SLURM interface.
GAIA contains two interfaces: an AWS ParallelCluster interface and a simulation interface. The AWS ParallelCluster is used to demonstrate our policies in a real testbed, while the simulation is used for large-scale experiments. The artifact also includes a configurable MPI program that illustrates different job lengths and resource requirements in the AWS ParallelCluster (Slurm) testbed.  
The artifact details instructions for executing GAIA in both environments and lists instructions for repeating Figures 6-10 in the simulation environment. We provided instructions for running the simulations and a Jupyter notebook to plot the results.