In this white paper, we discuss the Tungsten Replicator. Tungsten Replicator is a high-performance database replication tool that can extract data from various MySQL sources and apply that data to databases, data warehouses, NoSQL databases and messaging platforms. Tungsten Replicator is a complex piece of software that is based upon a few key and core design principles, optimized for continuity and speed. We hope you enjoy learning about how it works.
Table of Contents
- Introduction
- Basic Architecture
- Transactions
- Starting, Stopping, Restarting
- Replicator Roles
- Replicator Communication
- Replicator Management
- Components
- Extractors
- Internal Queuing Mechanism
- Appliers
- Filters
- THL Management and Storage
- Finite State Machine (FSM)
- THL and Data Management
- Transaction History Log (THL)
- THL and Transactional Headers
- Metadata
- Transactional Data
- Statement-Based Events
- Row-Based Events
- Row-Based Inserts
- Row-Based Updates
- Row-Based Deletes
- Immutability
- Storage
- THL Position
- Extraction From MySQL
- Architecture
- Reading the Binary Log
- Metadata and Environment
- Local and Remote Reading
- Binlog Management
- Applying to JDBC Targets
- Architecture
- Applying to MySQL
- Applying to Oracle & PostgreSQL
- Parallel Apply
- Applying to Data Warehouse Targets With Batch Loading
- Architecture
- Applying to Vertica
- Applying to Redshift
- Applying to Hadoop
- Applying to Native Targets
- Applying to MongoDB
- Applying to Kafka
- Filtering
- Architecture
- Filtering Points
- JavaScript Filters