At a glance
| Source connectors | Plugins for databases, files, Cloud Storage and other systems via the Hub |
| Destination connectors | BigQuery, Cloud Storage and other sinks through plugins |
| Batch or streaming | Both; streaming pipelines in all editions |
| Change data capture | Replication from MySQL, SQL Server and Oracle to BigQuery |
| Transformations | Visual transform plugins; custom plugins via the SDK |
| Orchestration | Triggers and schedules for pipelines |
| Deployment | Managed on Google Cloud; execution in your project on Managed Service for Apache Spark |
| Licensing | Per instance-hour by edition plus execution costs; built on open source CDAP |
Cloud Data Fusion suits Google Cloud teams that want a point-and-click pipeline designer, a plugin hub and lineage, with BigQuery as the usual destination. Instance charges run for as long as an instance exists, from USD 0.35 to USD 4.20 an hour, on top of Spark execution costs, so it fits steady integration programmes better than occasional jobs.
Overview
Cloud Data Fusion is Google Cloud's managed service for building ETL and ELT pipelines visually. Google describes it as powered by the open source project CDAP. You design pipelines in Studio, a web interface and REST API where pipelines are directed acyclic graphs of plugins (sources, transforms and sinks), and extend it with plugins and sample pipelines from the Hub. At run time Data Fusion provisions an ephemeral cluster of Managed Service for Apache Spark in your project and deletes it afterwards, or uses an existing cluster.
Instances come in Developer, Basic and Enterprise editions. All three include the visual designer, streaming pipelines, field and dataset level lineage and internal IP support; Enterprise adds role-based access control, and Basic and Enterprise have regional rather than zonal high availability. A separate Replication feature continuously copies changes from MySQL, SQL Server and Oracle into BigQuery. The latest release in the release notes when checked was 6.11.1.4 (15 July 2026); versions 6.9 and earlier are no longer supported.
Key features
- Studio visual pipeline designer with a REST API
- Hub of plugins, connectors and sample pipelines
- Replication (change data capture) from MySQL, SQL Server and Oracle into BigQuery
- Field and dataset level integration lineage
- Batch and streaming pipelines executed on ephemeral Spark clusters
- Developer SDK for building custom plugins
- Role-based access control in the Enterprise edition
Pros and cons
Pros
- Visual designer that does not require writing Spark code
- Built-in lineage at field and dataset level
- Published, region-independent instance prices and 120 free Basic hours a month
- Replication into BigQuery for common operational databases
Cons
- Instances are charged for as long as they exist, whether or not pipelines run
- Execution costs for Spark clusters, storage and BigQuery are billed on top of instance fees
- Centred on Google Cloud, with BigQuery as the only Replication target
- Concurrent pipeline execution is limited by instance version, and Google asks you to contact a representative for scalability details
Pricing
Cloud Data Fusion charges per instance for pipeline development, billed by the minute at hourly rates, from instance creation until deletion. As listed on the pricing page on 7 November 2026 (USD, which Google says is the same in all supported regions): Developer USD 0.35 an hour, Basic USD 1.80 an hour with the first 120 hours a month per account free, and Enterprise USD 4.20 an hour. Pipeline execution is billed separately at the rates of the Managed Service for Apache Spark clusters (named Dataproc in the Data Fusion release notes) that Data Fusion creates, plus any Cloud Storage, networking and BigQuery use.
Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.
Who it is best for
Google Cloud teams that want visual, governed pipelines into BigQuery, with lineage and some change data capture, without writing Spark code.
Who should look elsewhere
- Teams running occasional small jobs, where always-on instance charges outweigh the benefit.
- Multi-cloud or SaaS-heavy ingestion where a managed connector service fits better: see Fivetran alternatives.
Verdict
Cloud Data Fusion gives Google Cloud users a visual, CDAP-based pipeline tool with lineage and Replication into BigQuery. Its cost model rewards steady use, since instances bill per hour while they exist and Spark execution is extra. If your work is mostly streaming or code-first, also look at Google's Dataflow and Datastream, covered in the ETL tools guide.
Frequently asked questions
How much does Cloud Data Fusion cost?
As listed on 7 November 2026: Developer USD 0.35, Basic USD 1.80 (first 120 hours a month per account free) and Enterprise USD 4.20 per instance-hour, the same in all supported regions, billed by the minute. Pipeline execution on Spark clusters and other resources is billed separately.
Is Cloud Data Fusion open source?
The service is a paid Google Cloud product, but Google says it is powered by the open source CDAP project.
Does Cloud Data Fusion support change data capture?
Yes, through its Replication feature, which continuously copies data from MySQL, Microsoft SQL Server and Oracle into BigQuery.
Sources
- Cloud Data Fusion pricing
- Cloud Data Fusion overview
- Cloud Data Fusion Replication
- Cloud Data Fusion release notes
Checked November 2026.
How we research reviews and what star ratings mean: our editorial method.