burst lets you run your software remotely---on any sized virtual machine---without any change to your existing development process. All you need is a working Dockerfile.

We support remote computing on Amazon Web Services and will be adding more (Google Cloud Platform support is currently in beta).

Detailed installation instructions and usage examples can be found here: https://www.burstable.ai/

• Python3
• Docker version 19 or higher
• A folder/project with a working Dockerfile
• ssh keys
• AWS or Google Cloud Services account and access keys

Make sure you are running the necessary versions of python and Docker (need Python 3, Docker >= 19)

python --version
docker --version 

Note: if you want to contribute to the burst OSS project or just follow bleeding-edge development, install through gitHub as described here instead.

You must have a public/private ssh key pair, stored as ~/.ssh/id_rsa.pub and ~/.ssh/id_rsa by default.

If you do not already have ssh keys, run ssh-keygen to generate them:

ssh-keygen -t rsa -b 4096

Follow the instructions here: python-virtual-environments-a-primer

Launch the virtual environment and do the remainder of your installation and set-up inside the virtual environment.

Note that for development and testing on a local machine (ie your laptop), your local environment should be using the same version of Python as the Dockerfile for your project. Presently (as of release 1.1), the Dockerfiles in our examples use Ubuntu 20.04, which provides Python3.8. Differing Python versions may occasionally work, but specific package versions are not available for all versions of Python. This can lead to situations where a single requirements.txt file with pinned package versions will not work for both local development and burst. We strongly recommend setting up a virtualenv with the same Python version as provided by the base image in your Dockerfile.

pip install burstable

burst configure

Enter your configuration information as prompted to set up a remote compute service.

Download the tarball of examples from https://burstable.ai/examples

Unpack the examples tarball and cd into the hello_burst/ directory. You should see a file called hello_burst.py.

The Docker daemon must be running in the background to use burst. Test to make sure it is running and that you have the correct version:

docker --version

It should return something like:

Docker version 19.03.12, build 48a66213fe

burst build --gpu

This may take several minutes; be patient. After some output, you should see this:

---------------------OUTPUT-----------------------
Build phase 1 success
----------------------END-------------------------
Synchronizing folders
receiving incremental file list
Scheduling shutdown of VM at 52.27.54.55 for 900 seconds from now

Build phase 2 success
DONE

 burst run python3 hello_burst.py

The response should look like this:

---------------------OUTPUT-----------------------
Welcome burst-utioners! You're running a virtual machine with 4 cpus
The following GPU's are available:
Tesla T4
----------------------END-------------------------
burst: DONE

There are several burst examples posted here: https://burstable.ai/examples

We recommend walking through the CIFAR-10 example, which uses Pytorch to implement a Convolutional Neural Net (CNN) for image classification on the benchmark CIFAR-10 dataset. This example also illustrates how to use burst to run the CNN using a Jupyter notebook on a remote GPU, for real-time model building and manipulation.

The examples page also includes instructions and template files for setting up your own Python project to run through burst, and/or for using burst to run a Jupyter notebook on a remote GPU.