Skip to main content

Connect Astro to External Data Services


Before you can run pipelines on Astro with real data, you first need to make your data services accessible to your Data Plane and the Deployments running within it. This guide explains how to securely connect Astro to external data services using the following methods:

  • Public Endpoints
  • VPC Peering
  • Workload Identity (GCP only)

If you need to connect to a type of data service that requires a connectivity method that is not documented here, reach out to Astronomer support.

Public Endpoints

The fastest and easiest way to connect Astro to an external data service is by using the data service's publicly accessible endpoints.

These endpoints can be configured by:

Public connection traffic moves directly between your Data Plane and the external data service's API endpoint. This means that the data in this traffic never reaches the Control Plane, which is managed by Astronomer.

IP Allowlist

Some data services, including Snowflake and Databricks, provide an additional layer of security by requiring you to allowlist a specific IP address before you can access the service.

On Astro, each Cluster has a pair of unique external IP address that will persist throughout the lifetime of the Cluster. These IP addresses are assigned for network address translation, which means that they are responsible for all outbound traffic from your Astro Cluster to the internet. To retrieve the IP addresses for a given Cluster, open a ticket with Astronomer support and request it. If you have more than one Cluster, you will need to allowlist each Cluster individually on your data service provider.

VPC Peering

Each Cluster on Astro runs in a dedicated VPC. To set up private connectivity between an Astro VPC and another VPC, you can set up a VPC peering connection. Peered VPCs provide additional security by ensuring private connectivity, reduced network transit costs, and simplified network layouts.

To create a VPC peering connection between an Astro Cluster's VPC and a target VPC, reach out to Astronomer support and provide the following information:

  • Astro Cluster ID and Name
  • AWS Account ID or GCP Project ID of the target VPC
  • Region of the target VPC (AWS only)
  • VPC ID of the target VPC
  • CIDR of the target VPC

From there, Astronomer will initiate a peering request. To connect successfully, this peering request must be accepted by the owner of the target VPC.

Once peering is set up, the owner of the target VPC can expect to continue to work with our team to update the routing tables of both VPCs to direct traffic to each other.

DNS Considerations with VPC Peering (AWS Only)

To resolve DNS hostnames from your target VPC, your Cluster VPC has DNS Hostnames, DNS Resolutions, and Requester DNS Resolution enabled via AWS Peering Connection settings.

If your target VPC resolves DNS hostnames via DNS Hostnames and DNS Resolution, you must also enable the Accepter DNS Resolution setting. This allows the Data Plane to resolve the public DNS hostnames of the target VPC to its private IP addresses. To configure this option, see AWS Documentation.

If your target VPC resolves DNS hostnames using private hosted zones, then you must associate your Route53 private hosted zone with the Astronomer VPC using instructions provided in AWS Documentation. You can retrieve the ID of the Astronomer VPC by contacting Astronomer support.

Workload Identity (GCP Only)

Workload Identity is recommended by Google as the best way for data pipelines running on GCP to access Google Cloud services in a secure and manageable way. All Astro Clusters on GCP have Workload Identity enabled by default. Each Astro Deployment is associated with a Kubernetes service account that's created by Astronomer and is bound to an identity from your Google Cloud project's fixed workload identity pool.

To grant a Deployment on Astro access to GCP services such as BigQuery, you must:

  • Go to the Google Cloud project in which your external data service is hosted
  • Add the Kubernetes service account for your Astro Deployment to the principal of that Google Cloud project
  • Bind the service account to a role that has access to your external data service

Kubernetes service accounts for Astro Deployments are formatted as follows:


To find the namespace of your Deployment, go to your Deployment page in the Cloud UI and copy paste the value in the Namespace field.

For a Google Cloud project called astronomer-prod and a Deployment namespace called nuclear-science-2730, for example, the service account for the Deployment would be:

GCP has a 30 character limit for service account names. For Deployment namespaces which are longer than 24 characters, use only the first 24 characters when determining your service account name. For example, if your GCP project was called astronomer-prod and your Deployment namespace was nuclear-scintillation-2730, your service account would be:

For more information about configuring service accounts on GCP, see GCP documentation.