[Yandex Cloud documentation](../../../index.md) > [Yandex DataSphere](../../index.md) > [Step-by-step guides](../index.md) > DataSphere Inference > Creating a node

# Creating a node

You can deploy an individual notebook cell or a third-party [Docker image](../../../container-registry/concepts/docker-image.md) as a standalone service using [nodes](../../concepts/resource-model.md#resources).

{% note warning %}

When deploying and using models, you pay for the uptime of each node instance: from its start to deletion.

If you no longer need the service you deployed, delete the node.

{% endnote %}

If your project uses packages and libraries that are not included in the [list of pre-installed software](../../concepts/preinstalled-packages.md), first [configure your node environment](node-customization.md) using a Docker image.

## Node from a model {#from-model}


1. Select the project in your community or on the DataSphere [home page](https://datasphere.yandex.cloud) in the **Recent projects** tab.
1. In the top-right corner, click **Create resource**. In the pop-up window, select **Node**.
1. Enter a name for the node in the **Name** field.
1. Under **Type**, specify the resource type: **Model**.
1. In the **Models** field, select the saved [model](../../concepts/models/index.md) and specify the inputs, if required.

   {% note info %}

   When deploying [PyTorch models](../../concepts/models/index.md#supported-types), DataSphere cannot automatically figure out the input and output parameters.

   If you are setting the inputs, make sure to specify the outputs as well. When you set the inputs manually for any model type, DataSphere cannot automatically infer the outputs.

   {% endnote %}

   For XGBoost and LightGBM models, the names and types of input and output parameters remain unchanged:
   * Input parameters: `input__0` of the `TYPE_FP32` type, vector with the [N] length. For example, if N=32, specify `[32]` in the **Tensor dimension** field.
   * Output parameters: `output__0` of the `TYPE_FP32` type, scalar value. In the **Tensor dimension** field, specify `[1]`.

   {% note tip %}

   To create a node from XGBoost and LightGBM models saved in DataSphere before April 8, 2024, re-save them to the [model](../../concepts/models/index.md) resource.

   {% endnote %}

1. Under **Folder**, select the folder for the new resources.
1. Under **Provisioning**, select the [configuration](../../concepts/configurations.md) of [instance](../../concepts/deploy/index.md) computing resources, [availability zone](../../../overview/concepts/geo-scope.md), and the ID of the [subnet](../../../vpc/concepts/network.md#subnet) to host the instance. The subnet must be in the same availability zone where you created the community.
1. In the **Maintenance limit** field, specify the maximum number of node instances that can be shut down for maintenance at the same time.
1. Under **Access control list (ACL)**, click ![Add](../../../_assets/console-icons/plus.svg) **Add ACL** and specify the [IDs of the folders](../../../resource-manager/operations/folder/get-id.md) from which connections to the node are allowed. The folder ID of the user creating the node is specified by default.
1. Click **Create**.

To view all created nodes:
1. Select the project in your community or on the DataSphere [home page](https://datasphere.yandex.cloud) in the **Recent projects** tab.
1. Under **Project resources**, select **Node**.

## Node from a Docker image {#from-docker}

1. Select the project in your community or on the DataSphere [home page](https://datasphere.yandex.cloud) in the **Recent projects** tab.
1. In the top-right corner, click **Create resource**. In the pop-up window, select **Node**.
1. Enter a name for the node in the **Name** field.
1. Optionally, in the **Description** field, enter a description for the node.
1. Under **Type**, select the resource to base your node on: **Docker**.
1. Under **Docker image storage**, select the container registry. You can select Container Registry or any other registry. For Container Registry, specify:
   * **Docker image** in the following format: `cr.yandex/<registry_ID>/<image_ID>:<tag>`.
   * **Username** will automatically get the `json_key` value.
   * **Password secret**: In the project settings, select the [secret](../../concepts/secrets.md) containing the file with the [authorized key](../../../iam/concepts/authorization/key.md) for the service account from the list.

   To use an image from another Docker image storage, select the **Other** type and fill the remaining fields with the values required for connection to your registry.

1. Optionally, under **Docker image settings**, set the disk size available in the Docker image.

1. Under **Endpoint**:
   * **Type**: Select the node connection protocol, **HTTP** (**HTTP/2**) or **gRPC**.
   * **Port**: Specify the port for connecting to the node.
   * **Timeout**: Set the session timeout, in seconds.
   * **Idle timeout**: Set the idle time before disconnecting from the node, in seconds.
1. Optionally, under **Telemetry**:
   * **Type**: Select the telemetry service, **Prometheus** or **Yandex Monitoring**.
   * **HTTP path**: Specify the address to send telemetry data to.
   * **Port**: Specify the port to send telemetry data to.
1. Optionally, under **Healthcheck**:
   * **Type**: Node health check protocol, **HTTP** or **gRPC**.
   * **Port**: Port used for node health checks.
   * **Path**: Path to the resource to check.
   * **Timeout**: Check duration, in seconds.
   * **Interval**: Interval between checks, in seconds.
   * **Fails threshold**: Allowed number of failed checks.
   * **Passes threshold**: Required number of successful checks.
1. Optionally, under **Docker image settings**, set the total memory available in the Docker image.
1. Under **Folder**, select the folder for the new resources.
1. Under **Provisioning**:
   * **Instance configuration**: Select a [configuration](../../concepts/configurations.md) of the [instance](../../concepts/deploy/index.md) computing resources.
   * **Distribution by zones**: Add an [availability zone](../../../overview/concepts/geo-scope.md) and the ID of the [subnet](../../../vpc/concepts/network.md#subnet) to host the instance. The subnet must be in the same availability zone where you created the community.
   * **Maintenance limit**: Specify the maximum number of node instances that can be shut down for maintenance at the same time.
   * **Additional disk**: Optionally, add a secondary disk for the instance. If you selected multiple instances, a disk will be created for each one.
1. Under **Access control list (ACL)**, click ![Add](../../../_assets/console-icons/plus.svg) **Add ACL** and specify the [IDs of the folders](../../../resource-manager/operations/folder/get-id.md) from which connections to the node are allowed. The folder ID of the user creating the node is specified by default.
1. Click **Create**.

To view all created nodes:
1. Select the project in your community or on the DataSphere [home page](https://datasphere.yandex.cloud) in the **Recent projects** tab.
1. Under **Project resources**, select **Node**.

#### Useful links {#see-also}

* [Configuring the environment for deploying a standalone service](node-customization.md)
* [Updating a node](node-update.md)
* [Deleting a node](node-delete.md)
* [Creating an alias](alias-create.md)
* [Deploying a service based on a Docker image](../../tutorials/node-from-docker.md)
* [Deploying a service based on an ONNX model](../../tutorials/node-from-model.md)