Chapter 3. SSIS Lift and Shift

In this chapter, we will talk about the different services that can be used by ADF and SSIS in Azure. SSIS packages can now be integrated with ADF, and can be scheduled/orchestrated using ADF V2. The SSIS package execution capability makes all fine-grained transformation capabilities and SSIS connectors available from within ADF.

In this chapter, we will cover:

• SSIS in ADF
• Leveraging our package in ADF V2

SSIS in ADF

SQL Server Integration Services (SSIS) has been the Microsoft ETL predilection tool for more than a decade. A lot of enterprises have used SSIS to load their on-premises data warehouses since its inception in SQL Server 2005.

In the last couple of years, IT departments have had to deal with different kinds of data and specific toolsets to process them. SSIS has successfully been able to access cloud data from on-premises ETL servers since 2015 with the Azure Feature Pack (https://docs.microsoft.com/en-us/sql/integration-services/azure-feature-pack-for-integration-services-ssis?view=sql-server-2017). However, issues occur when most of the ETL is in the cloud and SSIS is in a small part of the chain. And, up until now, it was very complex to use ADF V1 as the orchestrator in the cloud, with some SSIS package calls in the pipeline.

The following sections will describe how SSIS on-premises can be successfully leveraged to interact with cloud data in ADF V2.

Sample setup

The first thing...

Leveraging our package in ADF V2

So far, we haven't done anything new, in the sense that everything we did was on-premises. This part of the book will focus on cloud leveraging of SSIS packages.

Before ADF V2, the only way to achieve orchestration with SSIS was to schedule our SSIS load on an on-premises (or an Azure) virtual machine, and then schedule an ADF V1.0 pipeline every n amount of minutes. If the data was not available at a specific time, the next ADF run would take it. Or, we had to tell ADF to wait for it before processing the rest of its pipeline.

Also, with the advent of SSIS 2017, the scaling out of package execution had to be done on-premises. There are a couple of issues with it:

• Who is responsible for the data warehouse data different usage? The developers that create and maintain the packages are not necessarily aware of the cloud implications of their processes. The data might be used in systems other than the ones they had in their specifications, when they first developed...

Summary

That was a pretty long journey, but we made it!

The most complex step is done now. In this chapter, we saw how SSIS can interact with Azure. We built services for integrating data using ADF V2. We're also able to refresh on-premises data from ADF in the cloud. In the next chapters, we'll add activities to this pipeline.