How-to | Deploy API services on Kubernetes#

If your Dataiku Cloud plan includes API services, you can deploy an API service on Kubernetes.

Important

Before deploying an API service on Kubernetes, your instance needs to have enabled the option to create custom code environments. If in the Code Envs panel of the Launchpad, you see the option to add a code env, your instance is ready. If not, click to enable custom code envs.

Note

IP allowlist isn’t compatible with this type of deployment at this time.

Create a Kubernetes cluster infrastructure#

First, you need to create the infrastructure:

  1. Navigate to the Local Deployer of you Design node.

  2. Click Deploying API Services.

  3. Go to the Infrastructures tab, and click + New Infrastructure.

  4. For the infrastructure type, select Kubernetes cluster. Then designate an ID and a lifecycle stage, and click Add.

At this point, you’ve created the infrastructure, with all fields of the form pre-filled. You can deploy API services on this infrastructure. Only members of groups added in the Permissions panel will be able to deploy on this infrastructure.

Tip

The default settings are intended to work for everyone. However, you can edit these settings for the whole infrastructure at any time. Alternatively, you can edit these settings at the deployment level to adapt them to each service.

You can also create several Kubernetes cluster infrastructures to host API services with different defaults settings or stages like test and production.

Quotas and activity tracking#

API quotas are based on your subscription. To monitor your API usage, go to the Launchpad’s Usage & Monitoring tab, and then find the API services tab to see your API resource quotas.

It’s possible the CPU and RAM used by your deployments is above the limit fixed in the settings. A container to retrieve your API queries logs is attached to each pod. This explains some overhead of resources.