Self-hosted Oracle CDC & replication · Founding design partners

Move Oracle data anywhere, and prove it landed.

A self-hosted change-data-capture & replication engine for Oracle — near-real-time (seconds-class), correctness you can verify, to fourteen targets. Your data never leaves your network.

Source

Oracle

capture · LogMiner

Engine

KinetiShift

apply · commit-ordered

Sinks

14 targets

Become a founding design partner See how it works

Pre-release evaluation software — runs on your own systems, with a convergence report at the end.

Where teams put it to work

Two wedges, one engine.

Oracle migration & consolidation

Keep the source database live while KinetiShift instantiates the target from a consistent SCN snapshot and applies changes on top — then cut over once kinetishift verify shows the two sides converged. Minimal downtime, correctness proven before you switch.

Oracle → lakehouse & analytics

Land Oracle changes continuously into Iceberg, Delta, Snowflake, BigQuery, Databricks or Parquet with faithful types — the modern-sink path the incumbents underserve. Near-real-time, no trail files, no intermediate hop.

What it is

Self-hosted Oracle change data capture, one binary.

Log-based capture built clean-room in Rust — a single static binary you run on your own infrastructure, including fully air-gapped. No stack to run.

Near-real-time (seconds-class)

Mines Oracle's online redo through the documented LogMiner interface — seconds behind the source, not minutes.

Correctness you can verify

kinetishift verify — a cryptographic, PK-ordered, cross-engine digest that proves source and target match.

Never a guessed value

Skip-loud by design: anything it can't decode with certainty is surfaced and counted, never silently written wrong.

Any-to-any reach

Six engines both directions plus eight analytics / lakehouse targets — one tool for migration, DR / failback, and feeds.

Your data stays put

Self-hosted; changes never route through a vendor cloud. Runs air-gapped for regulated estates.

Predictable pricing

A flat price per source Oracle database — no per-core tax, no volume metering, no surprises.

Connects

Six engines both directions. Eight analytics sinks.

Bidirectional

Oracle
PostgreSQL
MySQL
SQL Server
Kafka
MongoDB

Analytics sinks

Snowflake
BigQuery
Databricks
Apache Iceberg
Delta Lake
Parquet
ClickHouse
Elasticsearch

Why teams trust the data

Convergence you can check, not take on faith.

Proof, not promises

Built-in convergence verification plus a lag / cutover-readiness dashboard. You see the two sides match before you rely on it.

Clean-room engineering

Derived only from Oracle's own documented behavior. No copyleft dependencies — enforced in CI.

Yours to run

Single binary, bounded memory, Kubernetes / Helm ready. No Kafka cluster, no JVM, no managed middle-tier.

See how to verify a release

See it on your data.

Start with a scoped pilot on one of your Oracle databases — pre-release evaluation software, run on your own systems, with a convergence report at the end.

Start a pilot conversation