Cloud CDC Labs
Hands-on labs for implementing CDC with cloud-native and commercial platforms like AWS DMS, Snowflake, and Matillion.
Learn how to implement CDC pipelines using cloud-native and commercial platforms. These labs complement the open-source Debezium-focused labs with managed service alternatives.
These managed-platform labs were AI-drafted and then reviewed for CDC
accuracy against this site's standards — delivery is at-least-once
and correctness comes from idempotent sinks (upsert on the
primary key, ordered by log position, not ts_ms). Vendor
commands, versions, and pricing change often; verify against current
vendor docs before any production use.
AWS DMS
Build a fully managed CDC pipeline using AWS Database Migration Service to replicate changes from RDS to S3 and Redshift. Includes Terraform templates.
- ✓ Fully managed service
- ✓ No Kafka infrastructure needed
- ✓ Native AWS integration
Snowflake
Stream CDC data into Snowflake using Kafka Connect Sink Connector or Snowpipe for automated ingestion from cloud storage.
- ✓ Real-time data warehouse
- ✓ Native JSON support
- ✓ Auto-scaling compute
Matillion
Build visual CDC pipelines with Matillion's drag-and-drop ETL platform. No-code approach with timestamp-based and hash-based change detection.
- ✓ Visual pipeline designer
- ✓ Low-code/no-code
- ✓ Built-in scheduling
Oracle GoldenGate
Enterprise-grade CDC with bidirectional replication, durable checkpoint-based delivery, and conflict resolution. Supports heterogeneous replication across Oracle, MySQL, PostgreSQL, and SQL Server.
- ✓ Active-active replication
- ✓ Durable, checkpointed delivery
- ✓ Heterogeneous databases
Fivetran
Set up fully managed CDC pipelines with zero infrastructure. 500+ pre-built connectors with automatic schema drift handling and usage-based pricing.
- ✓ Zero infrastructure (SaaS)
- ✓ Auto schema evolution
- ✓ 500+ connectors
Choosing the Right Platform
Rough order-of-magnitude guidance, not benchmarks — latency, cost, and setup time vary widely with data volume, region, and configuration. Treat these as starting points and confirm against current vendor pricing and docs.
| Aspect | AWS DMS | Snowflake | Matillion | GoldenGate | Fivetran | Debezium |
|---|---|---|---|---|---|---|
| Deployment | Fully managed | Fully managed | EC2/VM instance | Self-hosted or OCI | SaaS (zero infra) | Self-hosted |
| CDC Method | Log-based | Stream/file ingest | Timestamp/hash | Log-based + polling | Log-based | Log-based |
| Latency | Seconds | Seconds-minutes | Minutes | 1-5 minutes | Sub-second | Sub-second |
| Targets | AWS services | Snowflake only | DW platforms | Heterogeneous DBs | 500+ destinations | Any (via Kafka) |
| Setup Time | 1-2 hours | 1-2 hours | 2-4 hours | 15-30 minutes | 4-8 hours | 4-8 hours |
| Complexity | Medium | Medium | Low | Very Low | High | High |
| Cost (monthly) | $26-500+ | $50-1000+ | $500-2000+ | $365-730+ (OCI) | $500-5000+ | $100-500 |
| Best For | AWS-native stacks | Analytics workloads | Business users | Enterprise Oracle | Rapid setup, SaaS-first | Event-driven apps |
Decision Guide
Choose AWS DMS if you:
- Already use AWS RDS or Aurora
- Target is S3, Redshift, or Kinesis
- Want minimal operational overhead
- Need built-in retry and error handling
- Prefer infrastructure-as-code (Terraform)
Choose Snowflake if you:
- Use Snowflake as primary data warehouse
- Need near-real-time analytics
- Have semi-structured data (JSON, Avro)
- Want separation of compute and storage
- Already have Kafka infrastructure
Choose Matillion if you:
- Prefer visual, drag-and-drop tools
- Have business analysts building pipelines
- Use Snowflake, Redshift, or BigQuery
- Need complex transformations beyond CDC
- Want integrated orchestration/scheduling
Choose GoldenGate if you:
- Have Oracle Database licenses or OCI subscription
- Need bidirectional (active-active) replication
- Require durable delivery and conflict resolution
- Zero-downtime migrations are critical
- Heterogeneous replication (Oracle↔MySQL↔PostgreSQL)
Choose Fivetran if you:
- Want zero infrastructure management
- Need rapid setup (15-30 minutes)
- Require broad connector library (500+)
- Prefer SaaS/cloud-first solutions
- Value automatic schema change handling
Choose Debezium if you:
- Need sub-second latency
- Building event-driven microservices
- Want full control over CDC logic
- Already have Kafka infrastructure
- Need custom transformations (SMTs)
Additional Resources
Version Support Matrix
Check current versions and compatibility for all CDC platforms.
View Version MatrixDatabase Quickstarts
Database-specific CDC setup guides for MySQL, PostgreSQL, Oracle, and SQL Server.
View Quickstarts