Single JVM Agent
No brokers, no Kafka, no ZooKeeper. One process handles capture, trail, and apply.
1 process · 0 dependencies
Heterogeneous Change Data Capture
One agent.Every database,replicated live.
Real-time replication across Tibero, Oracle, PostgreSQL, MySQL, MariaDB, and SQL Server — sub-second latency, one-to-many fan-out, and automatic self-healing. No Kafka, no ZooKeeper, no brokers.
<1s
END-TO-END LATENCY
1→N
HETEROGENEOUS FAN-OUT
~24MiB
SINGLE-TARBALL DEPLOY
0
DATA LOSS ON CRASH
Live Replication Topology
NeuronCDC sits at the center of your data
One JVM agent captures from any source and applies to any target concurrently. Every engine in your estate orbits a single point of control.
Tibero
S+T
Oracle
S+T
PostgreSQL
S+T
MySQL
S+T
MariaDB
S+T
SQL Server
S+T
SAP HANA
Coming soon
MongoDB
TARGET
Kafka
ROADMAP
NeuronCDC
Capture · Trail · Apply
Source & TargetTarget / RoadmapSame agent binary — only the YAML shape changes
Tip: click DB2 or Big Data (coming soon) to expand its own orbit
How It Works
Source to target, through one agent
Source System
DBMS
Trail files
Real-Time
Data Replication
NeuronCDC Agent
Target System
DBMS
Trail files
Replication Management
One JVM agent reads the source's redo/binlog/WAL and applies to the target over standard JDBC — monitored from a single web console.
Key Advantages
Built for heterogeneous production
Everything a change-data-capture platform needs, delivered as one obfuscated, license-protected Java agent.
Crash-Survivable Trail
A database-backed durable queue means no data loss on restart — the trail replays exactly where it stopped.
exactly-once on restart
Heterogeneous by Design
One source fans out to many targets across different vendors, in parallel, through standard JDBC.
1 → N, any vendor
Self-Healing Replication
Per-transaction auto-heal, a periodic validator, and a universal reconciler re-emit missing rows automatically.
:HEAL-C · :HEAL-V
Native Bridge Where Needed
C/JNA is used only for Tibero raw-redo decoding. Everywhere else it is pure JDBC.
libneuron_tblr.so
Enterprise-Grade Protection
Per-host license enforcement, ProGuard bytecode obfuscation, and SSH-key admin — no shared credentials.
.lic required at startup
How a row gets from source to target
Four stages, zero data loss
Capture reads at the lowest available layer — redo, binlog, or WAL. The checkpoint advances only after the target commits.
01
Capture
Reads redo, binlog, or WAL at the lowest available layer.
02
Trail
A crash-survivable, database-backed durable queue of events.
03
Apply
Batches and parallelizes JDBC execution against every target.
04
Checkpoint
Advances position only after the target commits — zero loss.
Source & Target Matrix
Any source, any target, via standard JDBC
Source and target type is just a YAML field. Cross-vendor DDL is replicated through a built-in type translator. Kafka is on the roadmap.
| Database | Capture mechanism | DDL sync |
|---|---|---|
| Tibero | libtblr via JNA (TAC/TAS raw device) | Yes |
| Oracle 12c–21c | LogMiner (incl. RAC + CDB/PDB) | Yes |
| PostgreSQL 10+ | Logical replication (test_decoding) | Yes |
| MySQL 5.7 / 8.x | Binlog (ROW format + GTID) | Yes |
| MariaDB 10.x / 11.x | Binlog (ROW format) | Yes |
| SQL Server 2012+ | Native CDC change tables | Yes |
Cross-vendor DDL is replicated through a built-in type translator. Kafka is on the roadmap.
What NeuronCDC Does
One platform, every replication need
From cross-vendor DDL to RAC-aware capture — the full feature surface, delivered by a single agent.
- Replication between different DBMS
- Replication across different operating systems
- Supports 1:1, 1:N, N:1, N:N , and cascade N:N:N configurations
- Selective replication by user, table, or entire dataset
- Flexible real-time sync: one-way, reverse, or bi-directional
- Supports replication in RAC and TAC clusters
- Protect your CDC with HA Agent against agent failure during CDC events
- Web UI for easy monitoring and management
- Offloads processing to minimize CPU usage on the source DB
- Data validation and correction capabilities
- Cost-effective alternative to foreign CDC and Initial Load
Flexible Replication Patterns
One binary, four topologies
All patterns share the same agent — the shape of your neuroncdc.yaml is the only thing that changes.
1:1
Basic replication
One source → one target. The simplest deployment.
1:N
Fan-out
One YAML, multiple targets — each can be a different vendor.
1:1:1
Cascade
Chain agents: Tibero → Oracle → PostgreSQL.
N:1
TAC / RAC fan-in
Per-node agent captures its own redo thread; all converge to one target.
Next Step
Get a quote tailored to your estate
Custom solution or a personal offer — tell us your sources, your targets, and your latency budget. We'll size the agent for you.
Contact PT. Viriya Teknologi Data