Data Factory, Moving multiple lookup worksheets from Excel to one lookup table in SQL Server

A current project has an xlsx containing around 40 lookups in individual worksheets

Each worksheet consists of a code and a description

We decide that we want every single lookup in one lookups table in SQL Server.

This will have a Lookup Name, Code and Description that we can then use for the rest of the project

We want to do everything in one go in Data Factory.

For this Im going to use a simple example with just 3 worksheets

Azure Data Lake Gen 2

We are going to store the source data within a data lake.

The Source data looks like this

Lookup B worksheet

Lookup C Worksheet

SQL Server

I have an Azure SQL Database and on it I create the one table that all the reference lookups will go into

CREATE TABLE [staging].[Lookups](
[LabelKey] [int] IDENTITY(1,1) NOT NULL,
[LabelName] varchar NULL,
[Code] [int] NULL,
[LabelDescr] varchar NULL,
[Importdate] [datetime] NULL
ALTER TABLE [staging].[Lookups] ADD DEFAULT (getdate()) FOR [Importdate]

LabelKey has been added just to create a valid key for the table. LabelName has also been added which will be the name of the worksheet.

Finally ImportDate is added because we want to know exactly what time this data was imported into the table

Now we need to provide Data Factory with a list of worksheets

CREATE TABLE [staging].[LookupNames](
[LabelKey] [int] IDENTITY(1,1) NOT NULL,
[Labels] varchar NULL,
[Importdate] [datetime] NULL
ALTER TABLE [staging].[LookupNames] ADD DEFAULT (getdate()) FOR [Importdate]

Lookup Names is our seed table and will provide us with the worksheet names

we have populated it like this

SELECT 'Lookup C' 

Data Factory

Linked Services

Firstly we need to provide our linked services. Source and destination

go to Linked services via

and choose new.

call it ADLS_LS and select your Azure Subscription and Storage account.

At this point the connection was tested and was successful so we didn’t need to do anything further

Next, create your Azure SQL Database Linked Service

And call is SQLDB_LS (Or what ever you feel is the right naming convention. _LS is good because you can see exactly what are the linked services in the JSON script created

Again add in your details (We used a role that we created in SQL Server DB specifically for data factory with GRANT EXEC, SELECT, INSERT, UPDATE, DELETE on all the schemas)

Ensure the connection is successful

Data Sets

Now to come up with the actual source and destination datasets. If we parameterise them then we can reuse a single data set for lots of other activities within the pipeline

Click on the and choose New dataset

Choose the Format. In this case its Excel

We don’t want to specify any of the location values until we get to the pipeline, including the worksheet

Make sure First row as header is ticked (Unless you don’t have a header in Excel)

And create parameters

This means we can use this one Data set for all the SQL data sources


Now to create the pipeline specifically for the lookup

This is the basic pipeline we are going to add.


First of all In Activities search for lookup and drag this into the pane

This uses the SQL dataset because we are going to use our SQL table that contains all the names of the worksheets.

Note that first row only is not ticked because we are bringing all the information through



We are going to get the entire data set (Value) fed into the GetLookups Lookup.

Sequential is ticked because we are going to move through all the worksheets names in the table (Ensure that your Worksheets have exactly the same name as what is specified in your table)

Click on the Activities (1) to get to the activity

Copy Activity within the Foreach

We now set up the source of the copy activity

We use all the parameters within the dataset and add in the information from our Azure data Lake Gen 2 in the Storage Resource.

Within our Lookups table there is a column called labelname and we are going to populate this with the Labels column from our item. Out Item in the foreach loop and was created via the Lookup. And that lookup contained all the columns from our LookupNames SQL Table

The data will go into the Lookups table

Thats everything. You should be able to test your Pipeline by clicking debug and the Foreach should move through worksheet specified within your lookupnames table and add your information into SQL

Truncating lookup tables before re adding data

we want to be able to repeat this process and unless we add a truncate statement into our process we will keep readding the same information

We can add the following Stored procedure into SQL

05/10/2020 Debbie Edwards - Peak - Truncate lookups
EXEC [staging].[USP_Truncatelookups]
Create PROCEDURE [staging].[USP_Truncatelookups]
IF EXISTS(SELECT * FROM [dbo].[sysobjects] WHERE Name = 'lookups')
TRUNCATE TABLE [staging].[Lookups]
DBCC CHECKIDENT ('Staging.Lookups', RESEED, 1)

And this can be added to the the Pipeline before the foreach loop and Lookup with a Stored Procedure Activity

You wont be able to see the Stored procedure if you havent granted EXEC access to the specific Database Role name and schema

Always give the least amount of privileges and them move up if you need to

--Bring back information about the members in your roles
SELECT AS DatabaseRoleName,
isnull (, 'No members') AS DatabaseUserName
FROM sys.database_role_members AS DRM
RIGHT OUTER JOIN sys.database_principals AS DP1
ON DRM.role_principal_id = DP1.principal_id
LEFT OUTER JOIN sys.database_principals AS DP2
ON DRM.member_principal_id = DP2.principal_id
WHERE DP1.type = 'R'
ObjectType = rp.type_desc,
PermissionType = pm.class_desc,
ObjectType = CASE
WHEN obj.type_desc IS NULL
OR obj.type_desc = 'SYSTEM_TABLE' THEN
ELSE obj.type_desc
s.Name as SchemaName,
[ObjectName] = Isnull(, Object_name(pm.major_id))
FROM sys.database_principals rp
INNER JOIN sys.database_permissions pm
ON pm.grantee_principal_id = rp.principal_id
LEFT JOIN sys.schemas ss
ON pm.major_id = ss.schema_id
LEFT JOIN sys.objects obj
ON pm.[major_id] = obj.[object_id]
LEFT JOIN sys.schemas s
ON s.schema_id = obj.schema_id
WHERE rp.type_desc = 'DATABASE_ROLE'
AND pm.class_desc <> 'DATABASE'
AND = 'db_NameofRole'

you should hopefully have a good pipeline to run in your lookup information into one lookup table and truncate that table when ever you run the process

Data Factory to move daily snapshot data in Azure SQL DB to files in a Gen2 Data Lake

I have the following problem to resolve (And this was my initial attempt at figuring out a solution)

I have a data source with data that contains a date. the data contains daily snapshot of a record. this means that a record will be in the data set once per day. This will amount to a lot of data and we would rather hold it as files in Azure Data Lake Gen2 Storage

The logic is to pull daily files from the source database into dated files within the Azure data Lake

Once running this will probably pick up a days data because the rest are already created. However on the initial run I want it to pick up all the days that have been loading.

At the minute I have about 6 months of data to load

Tables in Azure SQL Database (destination)

CREATE TABLE [audit].[MemberDailyMetricsDates](
     [DATEUTC] datetime2 NULL

This table collects all the dates from the source snapshot table

Remember, the table records the same record every day with any changes to the record.

CREATE TABLE [audit].[IdWatermarks](
     [TableName] nvarchar NOT NULL,
     [WatermarkValue] [bigint] NOT NULL,
     [WatermarkDate] [datetime] NULL,
     [WatermarkDate2] datetime2 NULL

this is where I add the dates from the tables to show where we are.

For example if we add the last lot of data from 02/01/2020 then this value will be stored in the watermark table. I record them in different formats just in case.

CREATE TABLE [audit].[ProcessingMetrics](
     [ID] [int] IDENTITY(1,1) NOT NULL,
     [TableName] varchar NULL,
     [DateProcessed] [datetime] NULL,
     [DateUTC] datetime2 NOT NULL,
     [Duration] [int] NOT NULL,
     [NoRows] [int] NULL,
     [ID] ASC
 ALTER TABLE [audit].[ProcessingMetrics] ADD  DEFAULT (getdate()) FOR [DateProcessed]

We can record meta data from the data factory Pipeline as it runs

Stored Procedures in SQL Database (destination)

/*Debbie Edwards
 17/12/2019 Created initial SP
 Create a watermark table
 EXEC [audit].[audit].[USP_Watermark]    '2014-12-31 00:00:00.000'

 ALTER PROCEDURE [audit].[USP_Watermark] @NullDate datetime2



 IF EXISTS (Select * FROM [audit].[IdWatermarks]  
 WHERE TableName = 'DailyMetrics')

 DELETE FROM [audit].[IdWatermarks]  
 WHERE TableName = 'DailyMetrics';

 --DECLARE @NullDate datetime2 SET @NullDate = '2014-12-31 00:00:00.000'

 WITH CTEDailyMetricsDates (DATEUTC)
 (SELECT ISNULL(MAX(DATEUTC),@NullDate) FROM [audit].[DailyMetricsDates]) 

 INSERT INTO [audit].[IdWatermarks] 
 (TableName, WatermarkValue, WatermarkDate, WatermarkDate2)
 FROM CTEDailyMetricsDates

This is the Stored procedure that you run to create the watermark

Currently I’m just running this for one table. You could redesign to run this SP for different tables.

Also note, If there is nothing in the table I am setting a default date to work from @NullDate

/*Debbie Edwards
 18/12/2019 Created initial SP
 Update Processing details
 EXEC [audit].[USP_UpdateMetrics]
 ALTER PROCEDURE [audit].[USP_UpdateMetrics] @TableName varchar(100),   @DateUTC datetime2, 
 @Duration int,  @NoRows int



 INSERT INTO [audit].[ProcessingMetrics]
 (TableName, DateUTC, Duration, NoRows)

 (@TableName, @DateUTC, @Duration, @NoRows)


the Pipeline will run this stored Procedure to add meta data to the ProcessingMetrics table

Data Lake Gen2 Storage Account

Along with the Destination SQL Database you need to have an Azure Data Lake

Quick Set up for the Azure Data Lake. Setting up an Azure Data Lake V2 to use with power BI dataflows in Service (As a data source) …. and Setting up a service principal will give you information on how to:

  • Set up the Storage Account
  • Ensure Hierarchical name space is enabled
  • Create a file system
  • Grant reader Role to power BI service
  • Set up the Service Principal (Set up the App Registration)

Set up Data Factory

Now we are ready to set up everything with a Data Factory

create your data Factory and then go to author and monitor


AzureSQLDatabasereportingdb is the Source Azure SQL Database

Our Destination is a Destination Azure SQL Database AzureSQLDatabase[destination]reportingdb

And we have a gen 2 Data Lake Storage Account AzureDataLakeStorageGen2 (Which will need the Service Principal account setting up in order to use) See setting up a service principal….


the Linked Services are in place. Now we can add Data sets

Source Data Set – Azure SQL database

Add an Azure SQL Server Dataset. the first thing we need are two parameters for the TableName and Table Schema

Connect this up to the Source Linked Service and use the parameters to create a table with schema

The Schema will change when ever you use different table parameters so no need to set at this point

Destination Data Set – Azure SQL database

Destination Data Set – Azure data Lake Gen2

First of all, create a FileName parameter. Because we are creating multiple files, each file needs to be renamed to the UTCDate from the Foreach loop later. We need to set this up within the For each rather than here. Therefore will just set the name as a parameter at this point.

What we want to do is add the file into our file system created in the data lake. and we want to add a date to every single file because these will be the snapshots

In add dynamic content


This is where the actual information for Filename from the data set will be added within the copy activity later.

I imported a schema from a test file I already created.

Pipeline 1 Create Member Daily Metric Dates into destination database

Our first part of the process is to create a list of dates in the source data that we dont yet have in the Data Lake. The first time we run it, thats literally every date we have so far in the source data base

Lets have a look at the Activities in the Pipeline


I know that my dates are in datetime2 format so I’m using this and changing it to WatermarkValue so all the steps always use WatermarkValue no matter what the format. Here is the query in full

SELECT MAX(WatermarkDate2) AS WatermarkValue From [audit].[IdWatermarks]
WHERE WatermarkDate2 IS NOT NULL
AND TableName = 'DailyMetrics'

We record the table name of the source in the Watermark table. Basically this will tell us what the date we need to use to work from. remember, we set up a default if there is nothing in the dataset to work with which will run everything into the data lake


This is where we get to the Copy part of the activity. We are simply copying the dates into the table of the snapshots that we haven’t done yet.

Here we add the parameters of the table we want to use from Source. These parameters were set up in the data set section.

Add a Query. We want the Distinct Date from the source table Where the DATEUTC (The column in the table) is greater than WatermarkValue from the Previous Watermark Activity

SELECT DISTINCT DateUTC FROM [dbo].[DailyMetrics]
WHERE DateUTC > ‘@{activity(‘LookupWatermarkOld’).output.firstRow.WaterMarkValue}’

Now we are at the destination data set

The only item in here is DateUTC


This will trigger the Watermark Stored procedure and uses the Azure SQL database destination Linked Service

This is everything required for Section 1 that will run all the dates into our table based on snapshots we haven’t yet prepared

Pipeline 2 CopyMemberDailyMetricsDateUTC

Again, lets have a look at these Activities in more detail


this is the Lookup Activity that takes the full record set of Dates from the destination SQL Server

Note that we haven’t ticked First row only because the entire data set is required

CopyDailyFilesIntoDL (ForEach Activity)

We are taking the output.value from our activity LookupDates

Because it only has one column we don’t need to specify anything further. output.value means the entire record set

Sequential – The Items will be read one by one from the data set

there are 2 activities within this foreach loop. Doubly click on the activity to see what is happening in the loop

Again Lets have a look at this in more detail


The source is the table in the source database because we want to store everything in daily snapshot files.

The Query is a SELECT FROM WHERE Statement. Select all the columns from the DailyMetric table where the DateUTC in the source data is Equal to DateUTC in @Item which is generated by the ForEach activity 

the sink uses the Data Lake. Remember, in the data set we set the Filename parameter and here is where is gets set. Click on the Filename to view Dynamic content

This is Concatenating a file name, the Item UTC Date and .csv

Use @item() to iterate over a single enumeration in ForEach activity . This value is generated by the ForEach activity (In the Pipeline)

We are setting the date as a string in order to use within the file name. Because the For each loop is setting the item for us and we are inside this activity we can create the filename in this copy activity rather than in the data set its self.

We can then simply Import schemas to view the mappings between the source and sink


Finally, we want to keep a record of everything we are going to be doing in the loop so we can run our stored Procedure to add the meta data into a table in our source database

We can use out SQL destination Linked Service for this

the Stored procedure contains 4 parameters.

We can take the Date from the Item in the foreach loop.

Duration, and NoRows can be added from metadata. See Monitor programmatically

  • Duration @activity(‘CopyDailyMetrics’).output.copyDuration
  • No Rows @activity(‘CopyDailyMetrics’).output.rowsRead

the table name is simple DailyMetrics


Finally, now that everything is complete we can truncate the date table. Outside of the foreachloop

the max date is held in Watermark which will be used to create the new files Next time (Set in Pipeline 1)

Test your solution

Now we have everything in place we can test each pipeline. By Clicking debug

You can see the files coming in via Azure Table Storage

If you get errors, a great way of debugging is to go into the code. I had an error and after a file nam it had /n/n.

I simply removed this empty row and it worked.

Pipeline 3

There is now a Pipeline 1 to create the dates. And pipeline 2 to create the files. This is good because they can be tested separately.

We need a top level Pipeline to run them

And this is the new model for the flow of this Data factory

Add a trigger

now we have successfully tested the Pipelines we can set them up in a trigger by adding a parent pipeline that runs all the Pipelines in order

Another Post will be created on Adding Triggers


You need to make sure that the last file (When you run it is complete) In our case the snapshot table is run at 12 AM and takes around 20 minutes so we need to set this pipeline off at 1 AM

Add Alerts to Monitor the Pipeline

Another Post will be created on Adding Monitors

Setting up a Service Principal for Azure Data Lake Gen 2 (Storage) to use with Data Factory

An Azure service principal is a security identity used by user-created apps, services, and automation tools to access specific Azure resources. Think of it as a ‘user identity’ (login and password or certificate) with a specific role, and tightly controlled permissions to access your resources

Azure Service Principal

I am constantly having to remind myself how to set up the Service Principal for Access to things like Azure Data Lake Gen 2 when I am setting up a data factory (Or using the storage with another web app).

So I wanted to write a blog post specifically on this.

As the example, imagine you are moving data from an Azure SQL Database to files in Azure Data Lake Gen 2 using Azure Data Factory.

You attempt to add a Data Lake Connection but you need a Service Principal account to get everything Authorised.

You need this so the Data Factory will be authorised to read and add data into your data lake

An application (E.g. data Factory) must be able to participate in a flow that requires authentication. It needs to establish Secure credentials. The default method for this is a client ID and a Secret Key.

There are two types of permissions

Application Permissions No user context is required. The App (E.g. data Factory) needs to access the Web API By its self

Delegated Permissions The Client Application (E.g. data Factory) needs to access the Web API as a Signed in User.

Create an App

In Azure choose App Registrations

Here you can create an app – New Registration

Provide a name for your app. e.g. DataFactoryDataLakeApp

Grant your Registered App permissions to Azure Storage

This will enable your app to authorise Requests to the Storage Account With Azure Active Directory (AD)

You can get to your app by going to Azure Active Directory

Then App Registrations and choose the App

In your new App, go to Overview and View API Permissions

Next go to Add a permission

Go to Azure Storage API which contains Data Lake Gen 2

Notice that we are setting up Delegated Permissions for Azure Storage

You are warned that Permissions have been changed and you need to wait a few minutes to grant admin consent.

I am not an admin so I always get my admin to go into Azure Active Directory and Grant Admin Consent for Peak Indicators

Note that your app now has configured permissions for Azure Active Directory Graph and Azure Storage

Assign your new app to a subscription

Now you have an app you need to assign Contributor status to it to the level of service you require in Azure, Subscription level, Resource group level or resource level.

For this app I am going to set it up against the subscription. First go to the Subscription you want to add it to and then Access Control (IAM)

I have added the app as a contributor

Creating a Key Vault

We will be selecting and creating IDs in the next steps, but instead of simply remembering your secret. Why not store it in a Key Vault.

  • Centralise Application Secrets
  • Store Secrets and Keys Securely
  • Monitor Access And Use

Lets set one up in our Proof of Concept area.

Create a Key vault if you don’t have one already

remember to add any tags you need before Review + Create

Once completed you can go to the resource (E.g. Data Factory) but for the time being that is all you need to do

Application ID and Tenant ID

You can now go into your new app in Azure (App registrations) to get more details for Data Factory (When you set up the connection)

Tenant from Data Factory will be mapped to Directory (Tenant ID) from the App Overview

Service Principal ID from Data Factory will be mapped to Application (Client) ID From the App Overview

Create a Client Secret

Next, create your Client Secret.

In your App go to Certificates and Secrets

Click New Client Secret

Im going to allow this secret to Expire in a year (Anything using the app will start to fail so you would need to set a new secret and re-authorise)

We can add this into the Key vault so we don’t lose it because once you have finished here you dont see it again.

Open a new Azure Window and Go to your new Key Vault

Go to Secrets

Click + Generate Import

Notice I have set the expiration date to match the expiry date of the app

Ensuring the Access is set for the Data Lake Storage

For this you need to have a Data Lake Gen 2 set up and Microsoft Azure Storage Explorer downloaded

In Microsoft Azure Storage Explorer, navigate to the storage

Then Right click on the File System (In this case factresellersales) go to Manage Access and add the app.

Notice that we have set Read Write and Execute for the app on the file system and all the files will inherit these permissions

Adding The Data Lake Gen 2 Connector in Data Factory (Test)

I have a Data Lake Gen 2 with some files and I want to move them into a SQL Data base.

To test, Open or create a Data Factory

Go into Author and Monitor. Then Author

Go to Connections, +New and Choose Azure Data Lake Gen 2

Tenant = Directory (Tenant ID) from the App Overview

Service Principal ID = Application (Client) ID From the App Overview

Service Principal Key (You can get it from Azure Key Vault. Click ON secrets,Then the name and current version

You can then copy the secret value and add it into Data Factory

Test your Connection

Create the Data Lake Data Set

Here is where you know that all your efforts all worthwhile.

Create a new Dataset which will be an Azure Datalake Gen 2

This is great. I have access to the files in the data lake. Achievement unlocked.

Create your website with
Get started