Automating Your Experimentation

Once you have developed your experiment and set everything up, you may want to automate that setup so you can easily re-run the experimetn later. There are multiple ways to automate your experiment.


You can write scripts using languages like Bash and Python to run commands on your nodes in a given scenario. For example, if you wanted to ping B from A and then ping A from B, in an experiment called test and project Share, you could write the following script in bash:

    ssh A.test.Share "ping -c 1 B"
    ssh B.test.Share "ping -c 1 A"

You can then save it, e.g., in a file called and run it by typing on


Similarly, in Python you can achieve the same with the following script:

    import paramiko
    ssh = paramiko.SSHClient()

    stdin, stdout, stderr = ssh.exec_command("ping -c 1 B")

    stdin, stdout, stderr = ssh.exec_command("ping -c 1 A")

You can then save it, e.g., in a file called and run it on one of your experiment nodes (not on by typing:



You can use DEW - distributed experiment workflows to design your experiment in a human-readable format and generate NS file and bash scripts. `

We provide more guidance in DEW YouTube channel as well as in documentation on DEW Web site.

If you use image Ubuntu-DEW on your nodes, all the commands you type and snippets of their outputs will be saved in your project directory. You can use the tool flight_log, which is automatically installed in that image, to remind yourself of the commands you ran in the past and to select those you want to include in a Bash script. The script will be automatically generated for you. More information about this direction is in DEW YouTube channel.


You can use our MAGI orchestrator to create scripts that will be more robust and readable than Bash scripts. Please see our orchestrator documentation for guidelines and examples.