If you are comfortable with the command line, Docker Machine is a more efficient way to provision a cloud Docker host than going through a web console. Docker Machine is already installed with Docker Desktop for macOS and Windows. If you're using Linux, or you don't want to install the full Docker Desktop, you can easily install a docker-machine library.
For the commands below to work, you will also need to configure docker-machine with the necessary credentials for your platform. Docker provides examples for how to do this for Amazon EC2, and we would recommend creating the credential file for most users
In the command below, we illustrate using
docker-machine to provision a remote burstable instance with reasonable specs on Amazon EC2. The t3.large instance has two burstable vCPUs and 8GiB of RAM - sufficient for most demo-type applications, and we use the most recent official Ubuntu 18.04 AMI, and provision in Virginia (us-east-1). For more serious applications, you will almost certainly want to increase the amount of disk storage beyond the default (e.g., the below will likely use more than half of the default 8GB EBS volume before you've loaded your own environments, code, and data!). You may also wish to use a more powerful instance type or deploy in a region that's closer to you, though note that for now, Gigantum datasets are local to us-east-1 (reducing cost and increasing speed of access). We use the name
gigantum-server below, though you are welcome to change the name! You may also wish to explore docker-machine drivers for other providers (note that not all providers have all features, for example only some have GPUs available).
docker-machine create --driver amazonec2 \ --amazonec2-ami ami-0ac019f4fcb7cb7e6 \ --amazonec2-instance-type t3.large \ --amazonec2-region us-east-1 \ gigantum-server
docker-machine create --driver amazonec2 ` --amazonec2-ami ami-0ac019f4fcb7cb7e6 ` --amazonec2-instance-type t3.large ` --amazonec2-region us-east-1 ` gigantum-server
You may wish to use a more powerful instance type or deploy in a region that's closer to you.
Updated about a year ago