New to Chameleon? Or are you a Chameleon veteran interested in exploring a new feature or tool you’ve seen on Chameleon, but not sure where to begin? Look no further - we’ve gathered the top Chameleon tutorials for you to use as inspiration! Ranging from full-length webinars to short overview videos, you can learn and explore new ways to experiment on Chameleon!

 

For those just beginning:

1. The Introduction to Chameleon (52 minutes)

If you’re new to Chameleon or teaching a class and looking to get students set up, this presentation and demo covers everything you need to know from what Chameleon is to launching your first instance.

Presented by Cong Wang, RENCI, September 2020

 

2. Introduction to Jupyter on Chameleon (1 hour 45 minutes)

Looking for an efficient way to run, present and share your research? Learn about using Jupyter Notebooks to run experiments on Chameleon while harnessing its ability to combine code, text and visualizations to enable the reproducibility of your experiments. 

Presented by Jason Anderson, University of Chicago, May 2019

 

3. 5-10 Minute Run-It-Yourself Videos

Also check out the last section of this blog, Reproducible Experiments, to see 5-10 minute videos with an accompanying Jupyter Notebook available on Trovi. Run the notebook yourself alongside the video, or use the videos in a class to help students become familiar with Jupyter Notebooks, Chameleon, Tensorflow and other machine learning libraries, tools and concepts!

 

For all things orchestration:

Now that you've deployed your first instance on Chameleon, perhaps you’re wondering how to repeatedly deploy something more ambitious, like a virtual cluster or a complex networking topology. It is very important to make the deployment of such complex experimental configurations repeatable so you don’t have to redo all of your hard work every time you want to run the experiment -- plus, conference submission rules may explicitly request that you provide a repeatable experiment. Orchestration automates experiment configuration setup so you only have to do the hard work once! See how it's done in the following videos. You can learn more about choosing the right orchestration in this blog post.

 

1. Orchestration: Using Jupyter Notebooks (30 minutes)

Interested in Orchestration but not sure where to get started? This video covers using Jupyter Notebooks to  orchestrate an experiment - from resource allocation to gathering and visualizing results and to tearing it all down. By using Jupyter Notebooks, you can conduct your experiment in a reproducible and easily documentable way! Automate configuring an experiment in multiple servers inside a custom network with Ansible.

Presented by Mauricio Tavares, RENCI, March 2020

 

2. Orchestration: Heat Templates (45 minutes)

This video builds on the previous Orchestration webinar and uses a heat template to create the entire experiment - reservation, servers, network, experiment - without human intervention (besides initially triggering the process!). Learn about what a heat template is, how to write one, and how to set up a reservation in the future that will automatically create your experiment from scratch at a predefined time. 

Presented by Mauricio Tavares, RENCI, May 2020

 

3. Automated Deployment (25 minutes)

This video builds on the previous webinars and highlights Chameleon’s Automatic Deployment mechanism to start a reservation and deploy an experiment without human intervention. This feature is useful if your experiment takes a long time to deploy -- some complex environments can take upwards of an hour -- then you can simply set it up to deploy automatically when your reservation starts. If your reservation starts at 7am and you’ve set up automatic deployment, you don’t need to get out of bed until it’s ready! 

Presented by Mauricio Tavares, RENCI, July 2020

 

For all things networking: 

Interested in deploying networking experiments across multiple Chameleon resources but not sure where to begin? The following tutorials can serve as jumping off points for networking experiments, but once you’ve mastered the basics, the Tips and Tricks section (look for posts published around summer 2020!) of the Chameleon blog offers a variety of networking tips, including Jupyter notebooks that you can run to practice setting up experiments.

 

1. Programmable Networks: Software Defined Networking (1 hour 10 minutes)

Learn about Chameleon’s Bring Your Own Controller (BYOC) functionality that enables tenants to create isolated network switches managed using OpenFlow controllers provided by the user. This webinar will demonstrate how to use Chameleon’s BYOC capabilities and provide tips and tricks to getting the most out of networking experiments.

Presented by Paul Ruth, RENCI, September 2019

 

2. Experiments Spanning Chameleon and GENI (Stitching) (1 hour)

Are you performing networking and distributed computing experiments but experiencing internet congestion or security issues that affect your network traffic? In this webinar, learn how to deploy Chameleon DirectStitch ports to connect isolated tenant networks at two Chameleon sites using a predefined Jupyter notebook.

Presented by Paul Ruth, RENCI, July 2019

 

For all things CHI@Edge, Chameleon’s new Edge Computing testbed:

After mastering Jupyter, orchestration and networking, what about expanding your experiment to range over edge devices and the cloud? Learn how to program your edge devices talking to the cloud from one Jupyter notebook. You can explore the other capabilities CHI@Edge has to offer or join the mailing list to stay up to date here. 

 

1. CHI@Edge (56 minutes)

Learn all about Chameleon’s new edge computing testbed, upcoming plans, IoT hardware and see a quick demo about how to use it.

Presented by Kate Keahey and Jason Anderson, University of Chicago, June 2021

 

For reproducible experiments you can run yourself: 

In need of quick inspiration? Check out these ~5 minute videos of reproducible experiments. All of the experiments are available on Trovi, so you can rerun them yourself alongside the video! These videos are also a great reference point for how to navigate Jupyter Notebooks if you’re unfamiliar with it. More detail about each of these videos including the links to the Jupyter notebooks is available in this blog post. 

1. Tiny-Tail Flash: Near Perfect Elimination of Garbage Collection Tail Latencies in NAND SSDs Reproduction (5 minutes)

2. LinnOS: Predictability on Unpredictable Flash Storage with a Light Neural Network (7 minutes)

3. Image Classification with AlexNet on Stanford Dogs Dataset (4 minutes)

4. Accuracy Levels with DAWNBench and TensorFlow End-to-End Training (6 minutes)

 

Feel free to explore the Chameleon YouTube channel for other tutorials and webinar recordings. If there’s anything specific you’d like to see, send us an email!