By deploying this terraform template via IBM Cloud Schematics or via Terraform, you accept the Terms and Conditions of the IBM License Agreement for Evaluation of Programs. If you do not agree to these terms, do not deploy this template.
This provisions a dedicated instance of H2O.ai Driverless AI in IBM Cloud utilizing IBM Cloud Schematics or with standalone Terraform. It follows http://docs.h2o.ai/driverless-ai/latest-stable/docs/userguide/install/ibm-docker.html#install-on-ibm-with-gpus.
Once created, its public IP address along with a username and password to log into the application will be displayed for easy access.
More specifically, it creates the following resources:
- a Virtual Private Cloud (VPC)
- a Subnet
- a Virtual Server Instance within the VPC and a particular region and availability zone (AZ)
- a floating IP (FIP) address on the public Internet
- a security group that allows ingress traffic on port 22 (for debug), on port 443 (SSL), and on port 12345 (H2O.ai DAI)
IMPORTANT: Reboots of the VM are not supported, and will result in loss of data. Back up any datasets or models prior to a reboot or shutdown of underlying VPC infrastructure.
NOTE: Please note that provisioning may take approximately twenty minutes.
To run as a standalone Terraform deployment, you need the following prerequisites.
terraform: v0.11.x or greater
ibm terraform provider: v0.24.x or greater
Use the IBM Cloud VPC Terraform Documentation for information on how to install Terraform and the IBM Terraform Provider.
You also need to have an IBM Cloud API Key.
- Clone this git respository
- Review the deployment attributes in the vm.tf file. You may use the defaults.
- Run
terraform apply
When deployment starts, it will ask you for your API key. The application will then take ~20 minutes to launch.
Simply run terraform destroy to remove the application infrastructure. The solution will also remove the VPC, Subnet, and all other associated resources it created. It will not touch other infrastructure in your IBM Cloud account.