ACM SIGCOMM 2023 Tutorial: Closed-Loop “ML for Networks” Pipelines

This is a repository with supporting materials for ACM SIGCOMM 2023 Tutorial: Closed-Loop “ML for Networks” Pipelines.

The repository structured as follows:

• requirements.txt - contains the list of required Python packages
• vm_prepare.sh and vm_prepare_2.sh - scripts to prepare the VM for the tutorial
• session_1: this folder contains presentation and demo materials for the first session of the tutorial (The Standard ML Pipeline: Problems and Challenges)
• session_2: this folder contains presentation materials for the second session of the tutorial (Beyond the Standard ML Pipeline)
• session_3: this folder contains demo materials for the third session of the tutorial
  • trustee_practice: this folder contains hands-on materials for Trustee practice of session 3
  • netunicorn_practice: contains hands-on materials for netUnicorn practice
    • notebooks: contains Jupyter notebooks for netUnicorn practice
    • scripts: contains preconfiguration scripts. These scripts are used by Docker Compose.
    • netunicorn-compose.yml - docker-compose file for netUnicorn practice
    • additional_materials: contains additional (optional) materials for the netUnicorn practice
• session_4: contains presentation materials for session 4 (Mini workshop)

Session 2: Trustee

• Web page: https://trusteeml.github.io/
• Trustee paper: https://dl.acm.org/doi/10.1145/3548606.3560609
• Extended version of the paper: https://github.com/TrusteeML/emperor/blob/main/docs/tech-report.pdf

Session 2: netUnicorn

• Web page: https://netunicorn.cs.ucsb.edu/
• netUnicorn paper: https://doi.org/10.1145/3576915.3623075
• Extended vesiorn of the paper: https://arxiv.org/abs/2306.08853

Session 3 Participant Instructions

If you are participating in the tutorial, please follow the instructions below to prepare your environment for the hands-on practice:

1. Get the IPv4 address of your virtual machine from one of the instructors.
2. Open the browser and navigate to the following URL: http://<VM_IPv4_address> (notice that it is an HTTP connection, not HTTPS).
3. You will see the Jupyter Lab password prompt. Enter the next password: sigcommtutorial.
4. Please, navigate to the session_3/netunicorn_practice/notebooks folder and open the session3.ipynb notebook.
5. Follow the instructor's instructions to complete the practice.

If you want to practice hands-on materials on your own, please follow the instructions below:

1. Clone this repository to your machine
2. Ensure you have python3 and pip3 installed on your machine (if you're using VMs, you can check vm_prepare.sh for instructions)
3. Execute instructions from vm_prepare_2.sh to install docker, pull needed images, download packages, and start jupyter lab (change jupyter lab starting folder if needed)
4. Open jupyter webpage and proceed with hands-on materials from corresponding sessions

Session 3 requires around 4GB of RAM on the host machine.