At a glance
| Sources | MySQL, MariaDB, Percona Server; Amazon RDS MySQL, Aurora, Azure and Google Cloud SQL |
| Targets | MySQL, PostgreSQL, Oracle, Amazon Redshift, Vertica, Kafka, MongoDB (including Atlas), Hadoop, Amazon S3 (CSV) |
| CDC mechanism | Reads the MySQL binary log into a Transaction History Log (THL); row-based binlog needed for heterogeneous targets |
| Initial load | Not documented as a built-in snapshot on the pages reviewed; heterogeneous target tables must start empty |
| Schema evolution | DDL is not replicated to heterogeneous targets; ddlscan generates target DDL |
| Latency | Not documented as a figure |
| Conflict handling | Manual tools to identify mismatches and skip failed transactions; active/active topologies documented |
| Monitoring | trepctl command-line status; Tungsten Dashboard is a separate Continuent product |
| Deployment | Linux tarball or RPM with tpm; AWS and Google Cloud marketplace images |
| Licensing | Commercial; quote from Continuent or hourly marketplace images |
Tungsten Replicator suits MySQL-family estates that need replication topologies native MySQL replication does not offer, such as fan-in, star or active/active, or feeds into Redshift, Vertica, Kafka or MongoDB. It is MySQL-source only, and pricing is by quote or by the hour through marketplace images, so compare it with open source CDC if you only need events in Kafka.
Overview
Tungsten Replicator extracts changes from MySQL, Amazon RDS MySQL, Amazon Aurora, Microsoft Azure and Google Cloud SQL sources and writes them into its own Transaction History Log (THL), giving each transaction a global ID. Appliers then write the changes to a target. The 8.0 manual documents appliers for MySQL, PostgreSQL, Oracle, Amazon Redshift (staged through S3), Vertica, Kafka, MongoDB (including Atlas), Hadoop and Amazon S3 as CSV. It supports statement-based and row-based replication; heterogeneous targets need row-based binlogs.
Continuent describes the replicator as the core foundation of Tungsten Cluster, its high availability product for MySQL, MariaDB and Percona Server, but it is also sold and documented as a standalone product. It supports parallel replication and topologies such as fan-in, star and active/active, and more than 50 built-in filters plus JavaScript filters for transforming data in flight. It is installed with the tpm tool from tarball or RPM packages, or launched from marketplace machine images.
Key features
- Binlog-based extraction from MySQL, MariaDB and Percona Server, including RDS, Aurora, Azure and Cloud SQL sources
- Transaction History Log with global transaction IDs, independent of native MySQL replication
- Appliers for MySQL, PostgreSQL, Oracle, Redshift, Vertica, Kafka, MongoDB, Hadoop and S3
- Topologies including primary/replica, fan-in, star and active/active, with parallel apply
- More than 50 built-in filters and JavaScript filters for in-flight transformation
- ddlscan tool to generate target DDL for heterogeneous appliers
- Installation with tpm (staging or INI configuration) or from AWS and Google Cloud marketplace images
Integrations and ecosystem
Pros and cons
Pros
- Mature MySQL-family replication with its own transaction log and global IDs
- Wide set of heterogeneous targets, from Kafka and MongoDB to Redshift and Vertica
- Topologies such as fan-in, star and active/active beyond native MySQL replication
- Same engine that underpins Tungsten Cluster, with a current 8.0 release line
Cons
- Sources are limited to MySQL, MariaDB and Percona Server
- No published standalone price; quote or hourly marketplace billing
- No Snowflake or modern lakehouse appliers documented in the 8.0 manual
- DDL is not carried to heterogeneous targets, so schema changes need manual work
Pricing
Continuent does not publish a Tungsten Replicator price: its pricing page covers only Tungsten Cluster (up to USD 8,400 per database node per year in 2026, with 24/7 support at 20 percent of software cost) and sends other products to a quote request. Continuent also lists Tungsten Replicator machine images on AWS Marketplace and Google Cloud, which its manual describes as available without support and which are billed per running instance hour; check the live listing for the target and instance type.
Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.
Who it is best for
MySQL, MariaDB or Percona Server teams that need flexible replication topologies or real-time feeds into Kafka, MongoDB, Redshift or Vertica, with vendor support.
Typical use cases
- Real-time feeds from MySQL into Amazon Redshift or Vertica for analytics
- Publishing MySQL row changes to Kafka topics as JSON messages
- Fan-in consolidation of several MySQL servers into one reporting server
- Replicating out of Amazon Aurora or RDS to another region, cloud or on-premises server
Who should look elsewhere
- Estates with PostgreSQL, SQL Server or Oracle sources: look at Debezium or Qlik Replicate.
- Multi-engine two-way sync with conflict rules across many nodes, where SymmetricDS is designed for the job.
Verdict
Tungsten Replicator is a specialised, supported option for MySQL-family sources with a broad list of targets and topologies native replication cannot match. Its MySQL-only sources, quote-based pricing and lack of DDL propagation to heterogeneous targets make it a narrower choice than general CDC platforms, so shortlist it when MySQL is the centre of your estate.
Frequently asked questions
Is Tungsten Replicator still available?
Yes. Continuent's documentation lists Tungsten Replicator 8.0.4, released 1 June 2026, as a current product next to Tungsten Clustering. Versions 6.1 and earlier are end of life.
Is Tungsten Replicator open source?
Older releases were published as open source, but Continuent's current pages present 7.x and 8.x as commercial software, sold through Continuent or as marketplace images. We found no official page offering current releases under an open source licence.
What is the difference between Tungsten Replicator and Tungsten Cluster?
Tungsten Replicator moves data from MySQL-family sources to MySQL or heterogeneous targets. Tungsten Cluster builds on the replicator to provide high availability, failover and multi-site MySQL clusters, and is priced per database node per year.
Can Tungsten Replicator write to Kafka?
Yes. The 8.0 manual documents a Kafka applier that turns row changes from the MySQL binlog into messages, alongside appliers for PostgreSQL, Oracle, Redshift, Vertica, MongoDB, Hadoop and S3.
Sources
- Continuent documentation index (current versions)
- Tungsten Replicator documentation
- Tungsten Replicator 8.0 manual
- Tungsten Replicator 8.0 introduction
- Continuent pricing (Tungsten Cluster)
- Continuent home page
Checked November 2026.
Listed in: Change Data Capture & Replication.
How we research reviews and what star ratings mean: our editorial method.