Azure Data Factory training looks different in 2026 than it did even 18 months ago. The DP-203 exam that covered ADF was retired in March 2025 and Microsoft now points new data integration learners toward Fabric, while the free Microsoft Learn modules remain the fastest way in. This guide maps the current learning path, the certification that replaced DP-203, what training costs and when upskilling your own team beats bringing in a consultant.
Table of Contents
What is Azure Data Factory and what does it do?
Azure Data Factory (ADF) is Microsoft’s fully managed, serverless data integration service. It moves and transforms data between on-premises systems, cloud platforms and SaaS applications through automated ETL and ELT pipelines, with more than 90 built-in connectors according to Microsoft’s introductory training module.
In practice, ADF is the orchestration layer of a Microsoft data platform. It does not store data and it does not analyse data. It collects data from source systems, applies transformation logic and lands the result where reporting and analytics tools can use it. That is why ADF skills rarely stand alone. Anyone learning it also touches the systems either side of the pipeline, typically SQL databases on the source side and a warehouse or Power BI model on the destination side.
The core concepts every learner must master
Six building blocks cover most of what a beginner needs, listed here in the order most people find easiest to learn them:
Pipelines are the workflows. Everything you schedule, run and monitor is a pipeline.
Activities are the steps inside a pipeline: copy data, run a stored procedure, call an API.
Linked services hold connection details for each external system, like connection strings with credentials.
Datasets describe the shape and location of the data an activity reads or writes.
Integration runtimes are the compute that does the work. The self-hosted runtime is the piece that securely bridges on-premises systems to Azure without opening inbound firewall rules.
Mapping data flows are the visual, no-code transformation designer, running on Spark compute that ADF manages for you.
Learners who build one real pipeline touching all six concepts retain them far better than those who watch a course end to end first. That is consistently what we see at OfficePro Consulting when clients ask us to hand over pipelines we have built: one guided walkthrough on their own data beats hours of generic video.
How do you learn Azure Data Factory in 2026?
The fastest route is Microsoft’s free, self-paced Learn modules, then a real project on your own data. Microsoft publishes introductory ADF training modules covering the basics, data integration with the integration runtime and large-scale ingestion and the full catalogue goes deeper. Paid courses add structure and a lab environment but the content largely mirrors the free path.
A working sequence we recommend to Australian teams:
- Complete the Introduction to Azure Data Factory module on Microsoft Learn.
- Work through the data integration module, paying attention to the self-hosted integration runtime if you have on-premises systems.
- Stand up a free-tier ADF instance and build one pipeline against a copy of your own data, not sample data.
- Add a mapping data flow with one join and one derived column, then break it deliberately and read the monitoring output.
- Only then decide whether you need a paid course, an instructor-led week or a certification path.
The order matters. Certification-first learning is the most common mistake we see. A certificate without a pipeline in production proves recall, not capability.
What skills do you need before learning Azure Data Factory?
SQL is the one genuine prerequisite. Most ADF work involves reading from or writing to relational databases and transformation logic frequently pushes down into SQL. Beyond that, basic familiarity with the Azure portal helps and a general grasp of what ETL means will make the documentation far easier to follow.
You do not need to code. Pipelines and mapping data flows are built visually and a capable analyst can be productive without Python or Spark knowledge. The deeper data engineering path is different: Microsoft’s current data engineering certification expects SQL, PySpark and KQL. For a team that just needs reliable data movement between systems, that bar is higher than necessary.
Which certification covers Azure Data Factory now that DP-203 is retired?
There is no longer a certification dedicated to classic ADF. Microsoft’s Azure Data Engineer Associate certification (DP-203), which listed Azure Data Factory among its core services, shows a retirement date of 31 March 2025 and is marked retired on Microsoft Learn. Its successor is DP-700: Fabric Data Engineer Associate, which tests data engineering on Microsoft Fabric, including pipeline orchestration and expects SQL, PySpark and KQL skills. The skills measured by the exam were updated again as of 21 July 2026, so check the current study guide before booking.
What this means in practice: if you want a current Microsoft credential that includes pipeline and orchestration skills, DP-700 is the path and it will teach you Fabric’s version of Data Factory rather than classic ADF. If your organisation runs classic ADF, hands-on capability matters more than the badge, because no current exam certifies it directly.
Should you learn Azure Data Factory or Fabric Data Factory?
Learn the one your organisation runs today, but know Microsoft’s direction. Microsoft describes Data Factory in Fabric as the next generation of Azure Data Factory and its documentation now advises those new to data integration to start with Fabric. Around 90% of ADF activities are already available in Fabric, so skills transfer well in either direction.
The table below summarises the practical differences a learner will notice between the two.
| Feature | Azure Data Factory | Data Factory in Fabric |
|---|---|---|
| Transformation designer | Mapping data flows | Dataflow Gen2 |
| Connections | Linked services plus datasets | Connections only, properties inline |
| On-premises access | Self-hosted integration runtime | On-premises data gateway |
| Compute management | You choose integration runtimes | Fabric manages compute |
| Publishing | Publish step required | Save and run |
| Certification | None current | DP-700 |
Our view at OfficePro Consulting: for an established ADF estate, retraining on Fabric is not urgent and private networking gaps in Fabric still matter for some regulated environments. For a team starting from zero with no existing pipelines, learning Fabric first is now the defensible default because that is where Microsoft’s investment and the only current certification sit.
How long does it take to learn Azure Data Factory?
A capable analyst with solid SQL can build and schedule a basic copy pipeline within a week of part-time study using the free modules. Genuine working proficiency, meaning parameterised pipelines, a self-hosted runtime, error handling and monitoring, typically takes one to three months of regular use. Microsoft’s instructor-led course for the current data engineering certification (DP-700T00) runs four days, which is a fair signal of how much ground the certified path covers beyond ADF basics.
The variable that moves the timeline most is not talent, it is whether the learner has a real pipeline to build. Skills decay fast without production use, which is a key input to the train-or-engage decision below.
What does Azure Data Factory training cost?
The self-paced path costs nothing but time. Microsoft Learn modules are free and a development ADF instance for practice costs very little under consumption billing. Paid options follow familiar models: online course marketplaces charge per course, instructor-led certification courses are priced per seat for a multi-day programme and the DP-700 exam itself carries a per-attempt fee. Pricing changes often enough that we deliberately avoid quoting figures, but the gap between the free path and a fully instructor-led certification path is significant and for a single in-house pipeline builder the free path plus real project time is usually the better spend.
The hidden cost is salary market pressure. Demand for data engineering skills in Australia remains strong, with specialist recruiter Big Wave Digital noting in its 2026 Australian data engineer salary guide that cloud platform depth including Azure is one of the factors that lifts pay. Training an employee into a scarce skill set raises retention stakes as well as capability.
When should you train staff and when should you engage a consultant?
Train your own people when data integration is recurring core work: multiple pipelines, regular changes and enough volume to keep skills warm. Engage outside help when the need is a small number of pipelines built once and maintained rarely, when nobody on staff has the SQL foundation or when the deadline is shorter than the learning curve.
The honest arithmetic for a mid-sized Australian organisation with two or three integration needs a year rarely favours a dedicated trained specialist. What works better in our experience is a hybrid: a consultant designs and builds the initial pipelines to a documented standard, then hands over to a trained internal owner who monitors and extends them. The internal person needs the one-week foundation, not the three-month proficiency, because the hard design decisions are already made. This is the model behind most of our data pipeline and analytics engagements and it pairs naturally with the broader question of how your business systems connect, covered in our systems integration guide.
If your integration work also spans SharePoint or line-of-business APIs, our guides to SharePoint API integration and automation and moving databases to Azure SQL cover the adjacent skills an ADF owner ends up needing and our API and integration services page describes how we run this work as engagements.
FAQ
Is Azure Data Factory training free?
The core path is. Microsoft Learn publishes free self-paced ADF modules covering the basics, integration runtimes and large-scale ingestion and a practice ADF instance costs very little under consumption billing. Paid courses and instructor-led programmes add structure and labs but cover largely the same ground. Budget time rather than money for the foundation.
Is Azure Data Factory being replaced by Microsoft Fabric?
Microsoft describes Data Factory in Fabric as the next generation of ADF and advises new data integration learners to start with Fabric, but classic ADF remains fully supported and widely deployed. Around 90% of ADF activities already exist in Fabric, so skills transfer. Existing ADF estates do not need urgent migration.
Is Azure Data Factory hard to learn?
Not for the basics. Pipelines are built visually and a person with solid SQL can schedule a working copy pipeline within about a week of part-time study. The difficulty rises with parameterisation, self-hosted runtimes and error handling, which is where real project experience matters more than course hours.
Do you need to know how to code to use ADF?
No. Pipelines and mapping data flows are designed visually without code. SQL knowledge helps significantly because most sources and destinations are relational. Coding skills like PySpark only become necessary on the deeper data engineering path, such as Microsoft’s DP-700 certification, which expects SQL, PySpark and KQL.
Is DP-203 still worth taking?
You cannot take it. Microsoft retired the DP-203 Azure Data Engineer Associate exam on 31 March 2025. The current data engineering certification is DP-700, Fabric Data Engineer Associate, which covers pipeline orchestration on Microsoft Fabric rather than classic ADF. Existing DP-203 holders keep the credential but cannot renew it.
Should we learn ADF or Databricks or SSIS?
They answer different questions. SSIS is legacy on-premises ETL and is where migrations come from, not where new skills go. ADF or Fabric Data Factory handles orchestration and data movement. Databricks is for heavy analytics and machine learning workloads. Most Australian mid-sized organisations need the orchestration skill first and Databricks only at genuine scale.
Getting started
Azure Data Factory training in 2026 comes down to a clear sequence: free Microsoft Learn modules, one real pipeline on your own data, then a deliberate decision between deeper training, DP-700 and outside help. The certification landscape has moved to Fabric but the underlying skill, moving data reliably between the systems your business runs on, has never been more in demand in Australia.
If you are weighing up whether to build this capability in-house or have the pipelines designed and handed over, we are happy to have a free initial discussion about what fits your situation. Get in touch with OfficePro Consulting.
