Oracle DBA

A Far Sync instance in Oracle Data Guard acts as a SYNC redo transport destination that facilitates zero data loss across long distances. By strategically placing the Far Sync instance at a metro distance from the primary database, it handles communication with a remote standby database located farther away. This setup ensures zero data loss in most scenarios, barring disasters that simultaneously affect the primary and Far Sync instance.

Here are the best practices for configuring a Far Sync instance, ensuring optimal performance and reliability:

1. Network Configuration

• Low Latency: The network between the primary database and Far Sync instance must maintain round-trip latency low enough to meet business requirements. Ideally, latency should be under 5ms for better performance, though successful configurations with higher latencies exist.
• Sufficient Bandwidth: Ensure the network can handle peak redo volumes, considering other traffic sharing the network. Use Redo Transport Compression to reduce bandwidth usage.
• Redundancy: Deploy redundant network links to ensure continuity during network component failures.
• Buffer Sizing: Follow standard Data Guard practices for TCP buffer sizes—set the send and receive buffer sizes to three times the bandwidth-delay product for optimal data transport

2. Storage Performance for Standby Redo Logs (SRLs)

• IOPS Capacity: Far Sync’s SRLs should be on storage capable of handling peak IOPS requirements, including those arising during redo gap resolution or peak database activity.
• Disk Performance: Disks at the Far Sync instance should match or exceed the primary database’s performance to prevent delays, archive log gaps, or increased round-trip times.
• Redo Log Groups: Configure SRLs with the same number of redo log groups as the primary database plus one additional group per thread, following MAA best practices.
• Manual Clearance of SRLs: For alternate Far Sync instances, clear SRLs manually before activation to ensure seamless SYNC transport and zero data loss.

3. SGA and Instance Configuration

• Minimal SGA Requirements: A small SGA suffices for Far Sync instances without impacting primary database performance.
• Optimal Configuration: Testing shows that an SGA size of 300MB with CPU_COUNT=1 (Linux) works efficiently.
• Instance Caging: Use Instance Caging to minimize SGA size and CPU usage during testing.

4. RMAN Archivelog Policy

• Configure the archivelog deletion policy at the Far Sync instance to:
  • SHIPPED TO STANDBY
  • APPLIED ON STANDBY This automatically manages disk space at the Far Sync instance, eliminating the need for backups of archivelogs at the Far Sync location, provided proper backups are maintained at the primary or standby.

5. Role Transition and High Availability

• Deploy Far Sync instances for both primary and standby databases to maintain zero data loss protection during role transitions.
• After a switchover, the Far Sync instance near the standby becomes active when the standby assumes the primary role.
• Protection Mode Compliance: Ensure the standby database has a Far Sync instance configured to avoid dropping the protection mode from Maximum Availability to Maximum Performance during planned role transitions.

6. Leveraging Fast Sync for Performance

• Fast Sync, introduced in Data Guard 12c, significantly improves primary database performance compared to traditional SYNC transport. Performance gains range between 4% and 12%, depending on network latency and I/O speed of Far Sync hardware.
• Virtualized environments showed no performance degradation compared to physical setups, provided CPU, I/O, and network requirements are met.

Conclusion

A Far Sync instance is an essential component for achieving zero data loss in long-distance Data Guard configurations. Proper network, storage, and configuration practices are critical for ensuring optimal performance and reliability. By adhering to these best practices, organizations can confidently deploy Far Sync to extend the capabilities of Oracle Data Guard in maintaining data protection and availability across distributed environments.