Shawn-Hx / TATA

DRL-based framework to solve task placement problem for data stream processing.

TATA

Throughput-Aware TAsk Placement in Heterogeneous Stream Processing with Deep Reinforcement Learning

Dependencies

• Python>=3.8
• Numpy
• scikit-learn
• PyTorch>=1.6.0
• pytorch_geometric

Data set

• DSP task graphs are in dsp_dataset
• Resource graphs are in resource_dataset

Quick start

python main.py

Flink integration

Please refer to Flink new scheduler.