Databricks Community

Arnold_Souza · ‎03-22-2023

We a facing a situation and I would like to understand from the Databricks side what is the best practice regarding that.

Question: Is it possible to have a cluster with a fixed Global IP on Databricks?

Details

We have a vendor that has a SQL Server database stored in Canada in another infrastructure that is not Azure. Their database is protected by a Firewall that limits the computers that can request access to it. All the workstations from the company resolve to the same Public IP. That works like that due to a Zscaler acting as a Flow Network Security. So on the Vendor side, it was easy to put this public IP in the allowed list of their firewall so that the connections could be established by the workstations.

The same behavior seems to not happen when we use Databrick’s Clusters. They seem to acquire a public IP that is dynamic. Therefore, the Vendor’s firewall does not recognize the Fixed IP (or Range of IPs).

So how to connect to an external database using a Databricks Cluster in this situation?

Anonymous · ‎03-28-2023

@Arnold Souza This is a common use case with the customers. You can use an Azure Firewall to create a VNet-injected workspace in which all clusters have a single IP outbound address. The single IP address can be used as an additional security layer with other Azure services and applications that allow access based on specific IP addresses. Please refer to the below KB article for more details.

https://kb.databricks.com/cloud/azure-vnet-single-ip

Arnold_Souza · ‎03-28-2023

We have the Databricks workspace Vnet injected. Unfortunately, We can't use a Nat gateway because it is raising an Error during the creation on Azure. The clusters that are managed by Databricks have "Basic" public IP by default and are not "Standard". So the Nat gateway is not supported on the container's public subnet. We do not have an Azure firewall or any NVA in the region where Databrick's workspace is placed.

We have raised a ticket to Databricks via Microsoft to get it solved. Without a proper answer since 27th March 23.

Otherwise, plan “B” is to recreate the workspace in a new subscription where we have Palo Alto Firewalls in place, which has a fixed outgoing IP.

Anonymous · ‎03-29-2023

Hi @Arnold Souza

Thank you for posting your question in our community! We are happy to assist you.

To help us provide you with the most accurate information, could you please take a moment to review the responses and select the one that best answers your question?

This will also help other community members who may have similar questions in the future.

Thank you for your participation and let us know if you need any further assistance!

Anonymous · ‎04-01-2023

@Arnold Souza If you file a support to Azure support they can help customize the Vnet by unlocking it as the Azure Databricks resources are deployed in a managed resource group. Your plan B also should be the way to go if option 1 does not work as expected. Once you deploy a new workspace you can migrate the existing artifacts as mentioned in the below document.

https://github.com/databrickslabs/migrate