RAC Migration Complete: GBase 8s Shared-Storage Cluster Passes Four Critical Tests to Take Over the Baton

Published on 2025-09-11

When the handover of RAC operations and maintenance became a top priority, Y City faced a critical challenge: finding a database solution that could pass four rigorous tests—achieving smooth migration to ensure business continuity, meeting high availability requirements with 24/7 instant response, handling hundreds or thousands of concurrent upstream requests, and strictly controlling costs. Several previous candidates had failed, until the arrival of GBase 8s Shared Storage Cluster (SSC), which finally turned this "relay race" around!

 

Test 1: Seamless Switchover!
Compatibility and Business Continuity Guaranteed

Objective

  • Full compatibility, supporting all RAC technical features  

  • Zero data loss and zero data corruption  

  • Zero downtime throughout the migration process, ensuring a smooth transition

Core Actions and Technical Support

1. Full + incremental synchronization with uninterrupted business operations: GBase 8s SSC first used the migration tool to connect to RAC and start receiving full data; after existing data migration was complete, it switched to the data synchronization tool, setting RAC as the source to capture new data in real time—all while RAC continued to support business without any interruption.  

2. O-compatibility for full-spectrum technical challenges: Faced with RAC's "compatibility arsenal" (data types, temporary tables, materialized views, system views, advanced system packages), SSC leveraged its O-compatibility feature to achieve full adaptation; for core PL/SQL capabilities (cursors, packages, nested tables, arrays, autonomous transactions), it also handled them seamlessly through compatibility, leaving no adaptation gaps.  

3. Dual-run verification for a more reliable switchover: After data synchronization stabilized, GBase 8s SSC executed the switchover command, seamlessly redirecting order data streams from RAC to itself; at the same time, it set up a reverse synchronization (SSC as source, RAC as target) for dual-run verification. After two weeks with no anomalies, RAC was formally retired, and the migration was completed successfully.

Results

"From data migration to business switchover, there was not a single minute of downtime, nor a single record mismatch—this compatibility and stability are outstanding!" —remarks from the project operations engineer  

 

Test 2: 24/7 Non-stop Operation!
Cluster Protection Delivers High Availability

Objective

  • Continuous business operation with zero risk of interruption  

  • Zero data corruption with automatic sub-second failover during faults  

  • Reduced manual intervention and lower operational burden  

Core Actions and Technical Support

1. Multi-node deployment for basic high availability: GBase 8s SSC deployed three nodes working together, with order data flows primarily handled by the first node, efficiently processing CRUD operations to ensure basic business operation.  

2. Cluster self-maintenance with sub-second fault response: When RAC simulated a "node failure" (terminating the first node process), SSC immediately triggered self-maintenance:  

  • CM (Connection Manager) switched business to the second node in sub-seconds, with no lag in order flows;  

  • The cluster daemon (cluster_alline) automatically started the "repair process" for the first node, completing service restart and rejoining the cluster within minutes, restoring redundancy.  

3. Repeated fault tolerance verification, rock-solid stability: Multiple node failures were simulated, and SSC consistently achieved sub-second failover and automatic repair, supporting business without interruption for seven consecutive days and nights, proving its high availability capabilities.

Results

"In the past, faults required manual node switching and on-site repair; now SSC handles it all automatically—finally, no more late-night monitoring!" —real feedback from the project DBA  

 

Test 3: 4,000 Concurrent Requests—No Problem!
Elastic Scaling Handles Massive Loads


Objective

  • No stalling under massive concurrent requests, with stable response times  

  • Zero transaction failures and guaranteed data consistency  

  • Dynamic resource adjustment based on pressure to avoid performance bottlenecks  

Core Actions and Technical Support

1. Connection redirection for load balancing: Facing the incoming order data flows, GBase 8s SSC used the connection redirection feature to distribute traffic across the three nodes, preventing any single node from being overloaded and significantly improving baseline concurrent processing.  

2. Dynamic elastic scaling to handle peak load: When requests doubled to 4,000 concurrent, SSC immediately triggered elastic scaling: it quickly added three new nodes, expanding from three streams to six, distributing the pressure across all nodes.  

3. Continuous stability verification: Under 4,000 concurrent requests for one hour, there were zero transaction failures, stable data processing, and no stalling or latency.

Results

"Previously, high concurrency always caused stalling; now even 4,000 concurrent requests are handled steadily, and we can add nodes anytime—performance ceiling is no longer a worry!" —endorsement from the project architect

 

Test 4: Halve Costs, Double Efficiency!
Clustered Deployment Cuts Costs and Boosts Efficiency

Objective

  • Meet all business requirements while reducing construction costs  

  • Lower operations investment and reduce long-term operational costs  

  • More efficient resource utilization, avoiding waste  

Core Actions and Technical Support

1. Deploy on existing hardware, simplify architecture to cut costs: GBase 8s SSC adopted a distributed architecture with three nodes each deployed on a single server, each equipped with the cluster_alline daemon and connection redirection; it reused existing shared storage, switches, and other equipment, adding only a few auxiliary materials, without needing to purchase expensive hardware.  

2. Automated operations to reduce labor costs: With automatic fault recovery, automatic load balancing, and other capabilities, SSC drastically reduced the need for manual intervention, significantly lowering long-term operations labor investment.

Results

"According to our calculations, the construction + operational costs of GBase 8s SSC are over 50% lower than the original solution, achieving the goal of halving costs without sacrificing performance." —recognition from the project manager  

 

Summary: Key Results of SSC’s Four Tests

In fact, SSC’s reliability has long been proven:

  • In the Shenyang Railway ANCC scenario, working with HAC, it supported an average daily passenger flow of 3 million and 7.5 million daily transactions;  

  • In the Shaanxi D5000 project, it easily handled efficient processing of 1,440 large tables;  

  • In the Bank of Kunlun FIDO system evaluation, it met all targets: sub-second automatic failover, 24/7 operation, full ecosystem adaptation, fast queries on tables with tens of millions of rows, high concurrency resistance, etc.

Now, GBase 8s SSC has once again proven with real-world examples that it can meet and excel in key dimensions such as compatibility, high availability, high concurrency, and cost control. Going forward, GBase 8s will continue to leverage its shared storage cluster to deliver more stable, efficient, and cost-effective database solutions for enterprises in more business scenarios!