Dynamic Registry Provisioning

This repository contains the material related to my Master's thesis.

The aim of this work is to dynamically provision and deprovision P2P registries (cache-based registries) in edge computing infrastructures to minimize the latency between users and applications, the provisioning time of application, and the infrastructure resource usage.

Configuring the environment

This project was configured with Poetry to manage its dependencies. Please ensure you have Python 3.10+ and Poetry installed. If not, you can install Poetry following the instructions here.

After installing these tools, you can install the dependencies of the project with the following command:

poetry install

Building the datasets

To build the necessary datasets to replicate the paper's experiments, navigate to the repository root folder and run the following command:

sh create_datasets.sh

Running the experiments

To run the experiments, navigate to the repository root folder and run the following command:

poetry run python run_experiments.py

Although the script runs a few cases in parallel, we recommend to run the script in background because it might take a while to complete all the executions.

Getting the results

After the executions finished, the logs are available at the logs directory. To analyze the results, run the analysis.ipynb notebook. You can modify this notebook to explore different metrics and datasets for various scenarios.