storing azure credentials best practice for azuremlsdk

Question

Looking at this the following creates a config.json file (I think):

new_ws <- create_workspace(name = <workspace_name>, 
                           subscription_id = <subscription_id>, 
               resource_group = <resource_group_name>, 
               location = location, 
               create_resource_group = FALSE)
write_workspace_config(new_ws)

Once this is done, it can be picked up like so:

ws <- load_workspace_from_config()

I would think that it is not best practice to source control the config.json file. What is the best approach please (in Windows)?

Thanks.

Answer 1

First, sorry if anything here seams inappropriate for your question since I do not know R.

If this is a project that won't be distributed (i.e to customers and be downloaded) I would save this data on an Environment Variable on you localhost or server and have all developers create a var as well. This will allow you to store all credentials and parameters without committing them.

This approach basically requires you to change the code which loads the credentials from the config file so it queries you localhost variables for the credentials. I found a nice guide on how to do that in R, check it here!

If this is a software that'll be distributed into production I would take a look at Azure Key Vault. This will allow you to safely store your secrets and get them when needed, authenticating with the user's account on an Azure AD or AD B2C. There's a nice guide here.

Best,

Felipe