Data Tool Guides

Databases

10 Best SQL Server ETL Tools for 2026

Compare 10 SQL Server ETL tools on how they read changes (CDC, change tracking or cursor), whether deletes come through, and where each fits.

Team Airbyte

March 31, 2026

Summarize with AI:

TL;DR

Short answer: with SQL Server ETL tools, the decision that matters most is made in SQL Server, not in the tool. Enable change data capture or change tracking and a tool can read only what changed, deletes included; skip it and every sync is a cursor query that misses deletes or a full reload that strains the source. Ten tools, and how each reads SQL Server:

Airbyte: open-source SQL Server source that reads CDC change tables, self-hosted or managed, into any of 700+ connectors' destinations.
Fivetran: managed connector that reads either CDC or change tracking, billed on monthly active rows.
Azure Data Factory: Microsoft's managed pipeline service, with native SQL Server CDC and a self-hosted runtime for on-premises servers.
SQL Server Integration Services: included with your SQL Server licence, with CDC components, but Windows-bound and dated.
Debezium: open-source CDC engine that streams SQL Server change tables into Kafka.
Matillion: SQL Server CDC through its agent, with transformation pushed down into your warehouse.
Talend: integration with data quality attached, now Qlik Talend Cloud with no free tier.
Informatica PowerCenter: mature enterprise ETL with PowerExchange CDC, though 10.5 left standard support in March 2026.
Pentaho Data Integration: visual ETL over JDBC, suited to scheduled batch loads rather than replication.
CloverDX: commercial platform for complex, heavily scripted transformation work.
One SQL Server constraint: CDC change data is cleaned up after three days by default. If a pipeline is down for longer than the retention window, the changes it missed are gone and it has to resync from scratch, so raise retention to match your longest realistic outage.

SQL Server ETL tools move data out of Microsoft SQL Server, Azure SQL Database and Azure SQL Managed Instance into warehouses, lakes and other systems. SQL Server is one of the most common transactional databases behind business applications, so it is often the first source an analytics pipeline needs, and the one where a careless full reload does the most damage to production.

What should you check in SQL Server before choosing a tool?

Your edition. CDC relies on SQL Server Agent jobs, so it is not available on Express edition. Standard, Enterprise and Developer support it, as do Azure SQL Managed Instance and most Azure SQL Database tiers.

Whether CDC or change tracking is enabled. Both are switched on per database and per table by someone with sysadmin or db_owner rights. Without either, tools fall back to cursor columns or full reloads.

Type mapping. DATETIME has no time zone and rounds to about three milliseconds, DATETIMEOFFSET does carry one, MONEY and DECIMAL need explicit precision downstream, and UNIQUEIDENTIFIER, XML and spatial types each need a deliberate destination type.

How do the SQL Server ETL tools compare?

Tool	How it reads changes	Deletes captured	Hosting	Pricing basis
Airbyte	CDC change tables, or a cursor column	Yes, with CDC	Cloud or self-hosted	Free self-hosted; Cloud capacity-based
Fivetran	CDC or change tracking	Yes	Cloud, hybrid	Monthly active rows
Azure Data Factory	Native CDC, or watermark queries	Yes, with CDC	Azure, plus self-hosted runtime	Per activity and data movement
SSIS	CDC source components	Yes, with CDC	On-premises Windows or Azure	Included with SQL Server
Debezium	CDC change tables, streamed	Yes	Self-hosted	Free, Apache 2.0
Matillion	CDC agent	Yes	Your cloud account	Credits
Talend	Queries; CDC via Qlik Replicate	With Qlik Replicate	Cloud or on-premises	Quoted subscription
Informatica PowerCenter	PowerExchange CDC	Yes	On-premises	Quoted licence
Pentaho Data Integration	JDBC queries	No	Self-hosted	Community and enterprise editions
CloverDX	JDBC queries	Only if you build it	Cloud or on-premises	Quoted licence

Which SQL Server ETL tools should you consider?

The first five read SQL Server's change data directly; the rest are batch or transformation tools that suit SQL Server when replication is not the main job.

1. Airbyte

Airbyte is an open-source data integration platform with 700+ connectors and a community of 25,000+ data engineers. Its SQL Server source reads CDC change tables, so inserts, updates and deletes all arrive downstream without repeatedly scanning the source, and falls back to a cursor column for tables without CDC. It loads into any Airbyte destination, including Snowflake, BigQuery, Databricks and Redshift.

What's unique about Airbyte?

Connectors are open source and editable, the no-code Connector Builder covers sources without a pre-built connector, and SQL Server usually sits alongside SaaS applications that Airbyte moves through the same pipeline. It runs as self-hosted open source, Airbyte Cloud, Self-Managed Enterprise, or Enterprise Flex, which keeps the data plane in your own environment when SQL Server holds regulated records.

2. Fivetran

Fivetran is a fully managed ELT platform with 700+ connectors. Its SQL Server connector can read either CDC or change tracking, applies schema changes automatically, and needs no infrastructure on your side. It bills on monthly active rows, so cost tracks how often your SQL Server rows change rather than how much data you hold.

Here is a closer look at Airbyte vs Fivetran.

3. Azure Data Factory

Azure Data Factory is Microsoft's managed data integration service and the cloud successor to SSIS for most new work. It reads SQL Server, Azure SQL Database and Managed Instance natively, supports CDC as well as watermark-based incremental copies, and reaches on-premises servers through a self-hosted integration runtime. It can also run existing SSIS packages, which eases migration. Pricing is per pipeline activity and data movement, and it is strongest when the destination is in Azure, such as Fabric or Synapse.

4. SQL Server Integration Services (SSIS)

SSIS ships with SQL Server, so there is no extra licence if your stack is already Microsoft. Packages are built in Visual Studio, and its CDC source and CDC splitter components read change tables and route inserts, updates and deletes. The trade-offs are a Windows-bound runtime, a dated development model, and few connectors for SaaS sources, which is why most new pipelines go to Azure Data Factory or a connector platform instead.

5. Debezium

Debezium is the open-source CDC engine many commercial tools build on. Its SQL Server connector reads CDC change tables and publishes every change as an event to Kafka through Kafka Connect, or to other sinks through Debezium Server. It is the free option for real-time SQL Server change streams, with the operational cost of running Kafka, managing offsets and building the load into your warehouse yourself.

6. Matillion

Matillion's Data Productivity Cloud is a cloud ELT platform with 100+ connectors and SQL Server CDC through its CDC agent. Its strength is pushdown transformation inside Snowflake, Databricks or Redshift, with dbt support since v1.70, and it bills in credits tied to job runtime. It makes most sense when SQL Server feeds a warehouse where Matillion already runs your transformations.

Here is a closer look at Airbyte vs Matillion.

7. Talend

Talend pairs integration with data quality and governance, with dedicated SQL Server components and log-based CDC available through Qlik Replicate. Two things to check before shortlisting it: Qlik acquired Talend in 2023 and now sells it as Qlik Talend Cloud, and Talend Open Studio, the free open-source edition, was retired on 31 January 2024, so there is no longer a free route in.

8. Informatica PowerCenter

Informatica PowerCenter is long-established enterprise ETL, strong on complex workflows across heterogeneous estates, with SQL Server CDC through PowerExchange. Two facts to weigh before choosing it: Salesforce completed its acquisition of Informatica in November 2025, and PowerCenter 10.5 left standard support in March 2026, so new investment is going into the cloud platform rather than PowerCenter.

9. Pentaho Data Integration

Pentaho Data Integration, long known as Kettle, is a visual ETL tool that reads SQL Server over JDBC and offers a large library of transformation steps. It suits scheduled batch loads and heavy reshaping in transit rather than replication, since it has no change capture of its own. Check the licensing of the edition you need, as the community and enterprise editions differ.

10. CloverDX

CloverDX is a commercial data integration platform built for complex transformation logic, with visual design plus scripting for the parts that need code. It reads SQL Server over JDBC, so incremental logic is yours to design. It fits teams whose SQL Server data needs substantial reshaping and validation on the way out, rather than straight replication.

Should you enable CDC or change tracking in SQL Server?

Both let an ETL tool read only what changed, and both are enabled per database and per table. They differ in what they keep:

Change data capture: SQL Server Agent jobs read the transaction log and write every change, with before and after values, into change tables. It gives a full history of each row between syncs, at the cost of extra storage and log-reader overhead, and it needs SQL Server Agent, so it is unavailable on Express.
Change tracking: records that a row changed, including deletes, but not the intermediate values. It is lighter on the server and simpler to run, and the tool reads the current row when it syncs. If a row changed three times between syncs, you get the latest version only.

For analytics replication, change tracking is usually enough and cheaper to run. Choose CDC when you need every intermediate change, for audit trails or event-driven use, or when your tool only supports CDC, as Airbyte's SQL Server source does. Either way, set retention longer than your longest plausible pipeline outage: the default for CDC is three days.

How do you start syncing SQL Server data with Airbyte?

Step 1: Set up SQL Server as a source

Create a read-only login with SELECT on the tables you want. For CDC, a DBA enables it on the database and each table and confirms SQL Server Agent is running. In Airbyte, open Sources, choose New source and select Microsoft SQL Server, then enter the host, port (1433 by default), database, schemas and credentials, choose CDC or a cursor as the update method, set encryption, and run the connection test.

Step 2: Set up a destination

Choose where the data lands: a cloud warehouse such as Snowflake, BigQuery or Redshift, a data lake, a database or cloud storage.

Step 3: Configure the connection

Select the tables to replicate, use incremental sync so each run reads only changes, and set a sync frequency well inside your CDC retention window.

The process is the same in Airbyte Open Source, which you can deploy yourself, and in Airbyte Cloud, which you can try free for 14 days.

Which SQL Server ETL tool should you choose?

Start in SQL Server: enable change tracking or CDC, and size the retention. Then, if SQL Server is one of several sources feeding a non-Azure warehouse, Airbyte replicates it alongside the rest, self-hosted or in Cloud, and Fivetran does the same with no upkeep if row-based pricing works for you. If everything lands in Azure, Azure Data Factory is the natural fit. Keep SSIS for existing packages rather than new pipelines, use Debezium if you want real-time events and will run Kafka, and reach for Matillion, Talend, Informatica, Pentaho or CloverDX when transformation drives the decision.

For tools that cover more than SQL Server, see our guide to the top data integration tools.

Replicate SQL Server with Airbyte

Move SQL Server tables into Snowflake, BigQuery, Databricks or any Airbyte destination with CDC, alongside the rest of your sources, self-hosted or in Airbyte Cloud.

Try Airbyte Cloud free for 14 days or talk to our team.

Frequently Asked Questions (FAQs)

1. What is the difference between SQL Server CDC and change tracking?

Change data capture records every change with its before and after values in change tables, using SQL Server Agent to read the transaction log. Change tracking only records that a row changed, so a sync reads the current version. CDC gives full history; change tracking is lighter and usually enough for analytics replication.

2. Does SQL Server Express support CDC?

No. CDC depends on SQL Server Agent, which Express edition does not include. Change tracking does work on Express, so it is the option for incremental extraction there.

3. How long does SQL Server keep CDC data?

Three days by default, after which the cleanup job removes it. If a pipeline is down longer than that, the changes it missed are gone and it needs a full resync. You can raise the retention period on the cleanup job to cover longer outages.

4. How do you move data from SQL Server to Snowflake or BigQuery?

For ongoing replication, enable CDC or change tracking and use an ETL tool with a SQL Server source and a Snowflake or BigQuery destination, such as Airbyte or Fivetran. For a one-off move, a bulk export with bcp to files in cloud storage, then a bulk load, is often quickest.

5. Is SSIS still worth using?

For existing packages, yes, and it costs nothing extra with SQL Server. For new pipelines, most teams choose Azure Data Factory, which can also run SSIS packages, or a connector platform with broader SaaS coverage and cloud destinations.

About the Author

Team Airbyte

Engineering Team of Airbyte.

10 Best SQL Server ETL Tools for 2026

About the Author

Related posts