Skip to main content
Open source · Oracle 9i → 26ai

Change data capture
with no agent on your database.

oracdc is a set of open-source Oracle CDC connectors for Apache Kafka Connect. RedoMiner reads redo logs directly — nothing installed on the database server, no extra CPU or I/O load.

Get StartedStar on GitHub
9i–26ai
Oracle versions
3 source
CDC connectors
50+
JMX metrics
0 agents
on the DB server
How it works

Redo logs in, Kafka topics out

RedoMiner parses Oracle redo directly and streams change events into Kafka Connect — then on to your sinks. The database does only lightweight dictionary queries.

Oracle Database
9i → 26ai · EE / SE / XE / Free
RAC · Data Guard · TDEredo logs
direct read
oracdc · RedoMiner
parse · transform · serialize
CDC · LOB · DDL · supplementalKafka Connect source
change events
Apache Kafka
topics · partitions
exactly-once deliverystreaming backbone
sink connect
Sinks
PostgreSQL · Oracle
JDBC sink connectorsdownstream targets
No software installed on the database server.  Zero production agent.
Why oracdc

Purpose-built for the Oracle ↔ Kafka boundary

Not a generic CDC tool with an Oracle plugin — connectors designed around how Oracle actually writes redo, across two decades of database versions.

Minimal database impact

RedoMiner reads redo logs directly without installing anything on the database server — eliminating the additional CPU and I/O load that agent-based and LogMiner approaches put on production.

Oracle 9i through 26ai

Every release from 9i to the latest 26ai — Enterprise, Standard, Express, Free and Personal editions all supported.

Multiple connectors

RedoMiner, LogMiner or SnapshotLog sources, plus JDBC sinks for PostgreSQL and Oracle.

50+ JMX metrics

Throughput, lag, transaction processing and resource use — full observability through standard JMX.

Flexible deployment

Docker, Confluent Hub, AWS Marketplace, Maven Central, or build from source — on any Kafka Connect cluster.

RAC & Data Guard

Full support for RAC clusters and Data Guard — including physical standby as a replication source.

Choose your connector

Three sources, one decision

Most teams should start with RedoMiner. The others exist for specific constraints — legacy access models or quick non-production snapshots.

Recommended

RedoMiner

Direct redo log reading
  • Minimal database load — dictionary queries only
  • Full LOB: BLOB, CLOB, NCLOB, XMLTYPE, JSON, VECTOR
  • Oracle RAC & DDL support
  • Minimal supplemental logging
  • TDE column encryption
Best for production at any scale.

LogMiner

Oracle LogMiner API
  • Moderate load — LogMiner sessions
  • LOB: BLOB, CLOB, NCLOB, XMLTYPE
  • RAC & DDL support
  • Requires ALL supplemental logging
For environments where direct redo access isn't available.

SnapshotLog

Materialized view logs
  • Higher load — not for production
  • No LOB, RAC or DDL support
  • No supplemental logging needed
For quick evaluation and non-production snapshots.
FeatureRedoMinerLogMinerSnapshotLog
MethodDirect redo readLogMiner APIMV logs
Database loadMinimalModerateHigh
LOB supportBLOB·CLOB·NCLOB·XML·JSON·VECTORBLOB·CLOB·NCLOB·XMLNo
Oracle RACYesYesNo
DDL supportYesYesNo
Supplemental loggingMinimal or ALLALL requiredNot required
TDE encryptionColumn encryptionNoNo
Get oracdc

Available everywhere you ship

GitHubsource & releasesConfluent Hubconnector pluginDocker Hubcontainer imagesAWS Marketplaceseller profileMaven Centralsolutions.a2.oracle
Commercial support

Production backing from the people who build it

Priority support, security hotfixes and early access to new features — directly from A2 Solutions, the maintainers of oracdc.

Contact A2 Solutions
Priority response SLAs
Security & hotfix backports
Early access to new features
Architecture & tuning guidance