Deploy a model on your own cloud resources

Requirements

🔓 You need a platform account with basic access level.

As every module available in the Marketplace is open-source, we offer the possibility to automatically deploy it in your custom cloud resources using the EGI Infrastructure Manager. This feature is available to any registered user (Access Policies).

✅ Pros

❌ Cons

  • You control where you deploy (no need to be a project member).

  • More work to configure the deployment,

  • You are consuming resources even when not actively making predictions (if not deployed as serverless).

For this, select a module in the Marketplace and click on Deploy ▼ IM.

../../_images/module.png

This will forward you to the Infrastructure Manager, with the module already selected.

../../_images/configure.png

If you are not logged in, you will be prompted to log in with your EGI Check-in account.

If you do not have any cloud provider configured, you will be prompted to configure them. You can do so by clicking on the Add button.

../../_images/configure_no_cloud.png

You can then select any of the cloud providers supported by the IM, and fill in the required information (e.g. Access and Secret Key in case of AWS).

../../_images/providers.png

Finally, you can select the cloud provider you want to use, and the base image you want to use for the deployment.

../../_images/configure_cloud.png

After clicking on Submit, the deployment will start. Then afther a few minutes, the VM will be deployed and the AI4EOSC module configured and you will be able to use it.

../../_images/inf_list.png

When the status changes to Configured, you can click on Outputs button to see the endpoint of DEEPaaS API UI of the module.

../../_images/outputs.png

Click on the link and you will see the API documentation, where you can test the module’s functionality, as well as perform other actions.

../../_images/deepaas_api.png