Reuse, less maintenance
Parameters collapse pipeline sprawl. Change a connection or a path once, in metadata, instead of in 60 places. New sources are rows, not rebuilds.
outcome: maintenance hours fall as the estate grows
orchestration layer · adf · fabric
Your pipelines and notebooks are the orchestra. Meet the conductor.
ETLMaestro is a metadata-driven orchestration tool that powers the ADF pipelines and Fabric notebooks you already run. It doesn't move data. It solves the master pipeline problem with a single, easy-to-use Power App for all of them: it executes every pipeline and notebook in checkpointed phases, retries automatically and notifies you when a task fails, and shows clear monitoring with estimated durations.
- sales_ods_nightly running succeeded
- erp_gl_extract succeeded
- claims_file_ingest failed · phase 4 retrying · checkpoint succeeded
- fabric_lake_merge running
- pbi_dataset_refresh queued running
- pricing
- From $200/mo, published. Every tier starts with a free two-month trial, no payment info needed. Tiers by configuration count on the pricing page, not behind a sales call.
- failure recovery
- Resume from the point of failure. A phase-4 failure resumes at phase 4, automatically. Minutes of rework instead of a full re-run.
- origin
- Version three. First built for a critical infrastructure client, then ChillETL, now ETLMaestro. 20 years of Microsoft ETL, 50+ projects.
the problem
What is a master pipeline?
A master pipeline is what happens when Data Factory development grows without a metadata approach: one sprawling parent pipeline calling children calling children, built a reasonable decision at a time. One person in the company can tell you what happened. Unless they quit.
Not scalable
Every new source is another rebuild on the canvas, and one changed path means edits in 60 places. Growth makes it worse, never better.
Tribal knowledge
The org chart says "data team." The architecture has a bus factor of one, and the restart order lives in nobody's documentation.
No visibility
"The daily job's still running" is not an answer. Without a status report, it's the only answer anyone has.
what changes
What does metadata-driven orchestration buy you?
Six things, each with a business outcome attached. If you've built a control-table framework by hand, this is that idea taken to its conclusion, maintained as a product instead of a side project.
Resilience
Failures resume from the point of failure with automatic retry. Work that already succeeded stays done.
outcome: a failure costs minutes, not a missed SLA
Visibility
The real-time status report shows what's running now, with ETAs drawn from historical run data. A manager reads it without translation.
outcome: "is the data fresh?" gets an actual answer
Scalability
Sequential and concurrent phases, unlimited schedules. The estate grows; the maze doesn't.
outcome: source count stops dictating headcount
De-risking
Tribal knowledge moves into the database. Any engineer who reads a data model can take over, so onboarding shrinks and the key-person dependency fades.
outcome: a resignation stops being an incident
Cost
Cheaper than Fabric's native monitor, which runs on your capacity units, and than Airflow hosting. No new platform: it runs on the Microsoft tools you already pay for.
outcome: from $200/month, published pricing
extendability
Works beyond Microsoft
The Azure Function integration lets ETLMaestro orchestrate anything with an API: Databricks, Snowflake, legacy schedulers, custom code. Publish a function app that implements three functions (start, check status, cancel) and that system joins the same schedule, the same retry policy, and the same status report as your ADF pipelines and Fabric notebooks.
One schedule for the whole estate
Mixed stacks are normal. The nightly run doesn't care which engine does the work; it cares what order things finish in.
Notifications on your channels
The notification phase is configurable: email, Teams, Jira, ServiceNow. The product doesn't lock you into one channel.
No bridge-burning
Adopting Databricks or Snowflake later doesn't strand your orchestration. They become one more thing ETLMaestro conducts.
Is ETLMaestro a fit for your team?
One honest qualifier before you book a call:
ETLMaestro is built for teams consolidating data from multiple sources into a central repository with ADF or Fabric. If that's not your architecture, this isn't for you yet, and we'd rather say so here than on a call.
plain answers
Questions engineers actually ask
Is ETLMaestro an ETL tool?
No. ETLMaestro does not move data. It's the orchestration and management layer on top of Azure Data Factory and Microsoft Fabric. A conductor doesn't play the instruments: your pipelines and notebooks do the data movement, and ETLMaestro decides what runs when, passes the parameters, and recovers what fails.
What is a master pipeline?
A master pipeline is what happens when Data Factory development grows without a metadata approach: one sprawling parent pipeline only one engineer can navigate. It isn't scalable, and the knowledge it encodes is tribal. When that engineer leaves, the knowledge leaves. Full breakdown here.
Is Airflow better than Data Factory?
That's a category error, like comparing a screwdriver to a hammer. Data Factory is the platform; Airflow is an orchestrator, which makes it ETLMaestro's actual competitor. If you're evaluating Airflow, compare totals: script-only with no UI included, roughly $200–2,000 a month in infrastructure for a production install, and you operate it. ETLMaestro includes a Power App UI and the status report, at lower total cost, on published pricing. The full Airflow comparison covers managed options and where Dagster and Prefect fit.
How is this different from Fabric's native monitor?
Fabric's monitor solves visibility only, and it runs on your capacity units to do it; we've watched organizations turn it off over the cost. ETLMaestro's status report doesn't run on your capacity, and it comes attached to the thing the monitor lacks: orchestration with resume-from-failure.
Couldn't we build this ourselves with a control table and Log Analytics?
You could, and plenty of teams have: a control table in Azure SQL, a Logic App for failure emails, a KQL workbook for the history. That covers logging. What the homegrown version usually doesn't cover is resume-from-failure, ETAs, one status report across the estate, and surviving the departure of the person who built it. If yours does all that, keep it.
What does "AI-assisted" mean here?
An optional assistant that drafts repetitive metadata SQL from plain English, working with Copilot in Microsoft environments. It reads only the metadata schema and never touches customer data. That's the whole feature.
What happens when a pipeline fails?
The status report shows the failed phase, and the automatic retry resumes from that phase rather than the beginning. Anything still failing is flagged on the channels you chose: email, Teams, Jira, or ServiceNow.
Find out how exposed you are
Ten questions, three minutes, no email. Your result names the archetype your estate matches and what to do about it.