GoldenGate: Basic Naming and Terminology's Part -2

1. Manager Process

The Manager process is the heart of GoldenGate.

• It must be running on both source and target systems
• Extract or Replicat processes cannot start without Manager
• Performs the following key functions:

Starts, stops, and monitors GoldenGate processes

Manages trail files

Controls ports and resources

Generates reports and error logs

Always ensure the Manager process is running before starting Extract or Replicat.

 

2. Extract Process

The Extract process runs on the source system and is responsible for capturing data changes.

Key Responsibilities

Captures transactional changes from the source database

Supports:

·      Initial Load (full data load)

·      Change Data Capture (CDC)

·      Can also capture DDL changes (where supported)

Data Capture Methods

Extract can capture data from:

• Database transaction logs (Oracle, SQL Server, DB2, MySQL, etc.)
• Vendor Access Modules (VAM) for databases like Teradata

3. Replicat Process

The Replicat process runs on the target system.

Key Responsibilities

Reads data from trail files

Applies:

Transactional data changes

DDL changes

Supports:

Initial Load

Continuous Change Synchronization

Replicat ensures that source and target databases stay in sync.

4. Collector Process

The Collector is a background process that runs on the target system.

Types of Collectors

• Dynamic Collector

Started automatically by the Manager

• Static Collector

Manually configured and started

Function

• Receives data sent by Extract or Data Pump via TCP/IP
• Writes received data into trail files
• These trail files are later read by the Replicat process

5. Trail Files

• Trails are GoldenGate’s temporary storage files.

Characteristics

• Stored on disk
• Written by Extract and read by Replicat
• Store captured transactional data in GoldenGate format

Types of Trails

• Local (Extract) Trail – Stored on the source system
• Remote Trail – Stored on the target system

Trail files provide:

• Fault tolerance
• Restart capability
• Data integrity

6. Data Pump Process

The Data Pump is an optional secondary Extract process on the source system.

How It Works

• Primary Extract → writes data to Local Trail
• Data Pump → reads Local Trail
• Sends data to Remote Trail on target

Benefits of Data Pump

• Protects against network or target failures
• Prevents Extract process abends

Supports:

• Data filtering
• Data transformation
• Consolidation from multiple sources

Without Data Pump

• Extract sends data directly to the target
• Data resides only in memory
• Higher risk during network or target outages

7. Data Source for Extract

When processing transactional data changes, Extract can read from:

• Database transaction logs (Oracle Redo, SQL Server Logs, etc.)
• Vendor Access Modules (VAM)

Used when direct log access is not supported

Example: Teradata

 

8.TRANDATA (Table-Level Supplemental Logging)

TRANDATA enables GoldenGate to capture enough information from the database redo logs to correctly replicate updates and deletes.

Why TRANDATA is Required?

By default, Oracle redo logs do not store full before-image data for all columns.
GoldenGate requires:

·      Primary key columns

·      Unique key columns

·      Or all columns (if no key exists)

TRANDATA ensures that this required information is logged.

What Happens Without TRANDATA?

• Updates may fail on target
• Deletes may not find matching rows
• Replication errors like:

9. Supplemental Logging (Database-Level Logging)

Supplemental logging is enabled at the database level to allow additional column data to be written into redo logs.

GoldenGate requires minimal supplemental logging. 

10. Supported Data Types

GoldenGate supports the following data types:

• NUMBER
• BINARY FLOAT
• BINARY DOUBLE
• CHAR
• VARCHAR2
• LONG
• NCHAR
• NVARCHAR2
• RAW
• LONG RAW
• DATE
• TIMESTAMP