This article demonstrates how to dynamically configure parametrized linked services and datasets during runtime, enabling the use of a single pipeline to access multiple sources/targets and avoiding the creation of multiple ADF entities.
Use Case
As an ISV with multi-tenant solution that pulls data from multiple similar sources through different connection strings, you might be able to leverage a single ADF/Synapse pipeline which would differ only in parameter values.
Target architecture
To minimize the operational overhead of on-boarding new customers, we will create a single pipeline with a parameterized linked service and dataset, configured during runtime, which will be used to access all the different storage accounts.
Create pipeline and activities
For this sample, we will pull the data from each customer storage and save it in the ISV storage. A Copy() activity will be defined inside a ForEach() activity that will iterate over an array of customers.
NOTE: To keep this sample more generic, the list of customers was hardcoded in a pipeline parameter. However, for production this is an anti-pattern and we strongly advise using a metadata table or file to dynamically pass these values in.
1. Create a pipeline.
2. Create a pipeline parameter called customers where the value is the Array of customers with all the information to access and store the customers files. The array of customers follows the bellow format:
customers = [
{
"account_name": "customerastorage",
"account_secret": "customer-a-storage-secret-1",
"container": "data",
"filename": "customer_a.csv",
"target_container": "customer_a"
},
{
"account_name": "customerbstorage",
"account_secret": "customer-b-storage-secret-1",
"container": "data",
"filename": "customer_b.csv",
"target_container": "customer_a"
}
]
3. Add a ForEach() activity to the pipeline and under Settings pass the customers Array, stored in @pipeline().parameters.customers, to Items.
4. Add a Copy() activity to the ForEach().
Create a parameterized linked service
1. Create a new linked service and add the parameters account_name and account_secret, that will be populated during runtime to access each customer storage.
2. Select the 'Enter manually' method.
3. Click on the URL text box and select the 'Add Dynamic Content' option that will be displayed bellow the text box. Enter the account URL manually and replace <accountname> with the linked service parameter account_name.
4. Select the Key Vault that stores the secrets of the customers accounts and assign the linked service parameter account_secret to Secret name.
5. In case you want to parameterize the linked service to the target account, create a new linked service and repeat the previous steps.
NOTE: Do not parameterize passwords or secrets. Store all secrets in Azure Key Vault instead, parameterize the Secret Name.
Create the datasets
Input Dataset
1. Create a dataset.
2. Select the parameterized linked service to access the different accounts.
3. Define the linked service properties/parameters. In this sample, account and Secret will be retrieved from the current loop @item() during runtime.
4. Define the path to access the files. Container, Directory and Filename will be retrieved from the current @item() during runtime.
5. Define this dataset as Source dataset of the Copy() activity.
Output dataset
1. Create a dataset.
2. Select the linked service to access the ISV storage.
3. Specify the path where the files should be saved. In this sample, each customer files will be saved in a different container. The values for the target container, directory and filename will be retrieved from the current loop @item() during runtime.
4. Define this dataset as Sink dataset of the Copy() activity.
FastTrack for Azure: Move to Azure efficiently with customized guidance from Azure engineering. FastTrack for Azure – Benefits and FAQ | Microsoft Azure
