Cassandra to Spark
real-time replication
TwinCommit replicates Apache Cassandra to Apache Spark in real time by reading the Cassandra transaction log and applying every committed change exactly once and in commit order. Use it to process every committed Cassandra change in real time with Spark, with no triggers or capture tables on the source.
Cassandra
log reader
commit-ordered · exactly-once
Spark
native writer
How capture works on Cassandra
TwinCommit tails the CDC commit-log segments on every node, deduplicates replica writes, and orders mutations by write timestamp per partition.
Prerequisites
- ✓cdc_enabled: true in cassandra.yaml
- ✓WITH cdc = true on replicated tables
- ✓Agent sidecar or shared volume access to cdc_raw on each node
How apply works on Spark
The TwinCommit Spark connector exposes each pipeline as a Structured Streaming source with offsets mapped to log positions, so your jobs get exactly-once semantics with Spark checkpoints.
Requirements
- ✓The twincommit-spark connector JAR on the driver class path
- ✓Network access from the driver to the TwinCommit change feed endpoint
Setup
Set up Cassandra → Spark in four steps
- 1
Prepare Cassandra
Run twincommit check. It verifies cdc_enabled: true in cassandra.yaml and the other prerequisites, and prints the exact fix for anything missing.
- 2
Point at Spark
Add a target block for Spark. TwinCommit can generate target DDL with twincommit ddl -apply.
- 3
Snapshot
twincommit snapshot copies each table under a brief share lock and records the exact log position.
- 4
Stream
twincommit run streams every committed change from that position, exactly once and in commit order.
source:
type: cassandra
contact_points: [cass-1, cass-2, cass-3]
cdc_dir: /var/lib/cassandra/cdc_raw
target:
type: spark
listen: 0.0.0.0:7443
tls: true
tables:
- source: sales.orders
- source: sales.customersCassandra to Spark replication FAQ
How does TwinCommit replicate Cassandra to Spark?
+
TwinCommit reads changes from the Cassandra transaction log, buffers them per transaction, and applies each committed transaction to Spark in commit order. The TwinCommit Spark connector exposes each pipeline as a Structured Streaming source with offsets mapped to log positions, so your jobs get exactly-once semantics with Spark checkpoints.
Does Cassandra to Spark replication need triggers or capture tables?
+
No. TwinCommit reads the Cassandra log directly and creates no objects in the source database.
Is Cassandra to Spark replication exactly-once?
+
Yes. TwinCommit commits its checkpoint in the same Spark transaction as the data, so a crash or restart never loses or duplicates a transaction.
What Cassandra versions are supported?
+
Apache Cassandra 4.0, 4.1, 5.0 · DataStax Enterprise 6.8+ · Astra DB (target) · ScyllaDB (target).
Stream Cassandra to Spark today.
Download the free Developer edition, or have one of our replication engineers walk you through a migration plan.