Consolidating Distributed Database Systems onto Oracle Exadata – Experiences from a Major Bank in Vietnam

In the journey of digital transformation and data infrastructure optimization, banks and financial institutions in Vietnam face a significant challenge: database systems that have evolved in a fragmented manner over multiple phases, spread across multiple platforms, including legacy Exadata racks, non-Oracle servers, and smaller systems serving specific business needs.

A major bank in Vietnam tackled this challenge by implementing a project to consolidate all 12 of its scattered database systems onto a modern Oracle Exadata platform. Executed by the MPS Vietnam team, this project not only delivered outstanding technical results but also provided valuable lessons in strategy, operations, and cost optimization.

Initial Challenges

Prior to the consolidation, the bank operated numerous Oracle database systems of varying sizes. Some ran on older Exadata systems, others on non-Oracle servers, and the rest were distributed across data centers serving specific business functions. This created a complex matrix of issues:

• Fragmented Management: The IT team had to allocate resources to manage each system separately, with distinct interfaces, tools, and processes.
• Resource Wastage: Hardware resources were dispersed, preventing flexible sharing and leading to inefficiencies.
• Unstable Performance: Older systems struggled with performance, especially for high-demand tasks.
• Costly Maintenance: Upgrades and maintenance were labor-intensive, time-consuming, and expensive.

The biggest challenge was the constrained budget, which required consolidating as many databases as possible ,onto the new platform in a single phase. Additionally, the initial design had to comply with PCI-DSS standards, mandating complete logical separation of systems to ensure security and compliance. This demanded precise planning from the outset, with foresight for future scalability and no room for errors.

Solution: Consolidation to Oracle Exadata Database Server

After a thorough assessment, the MPS team proposed a comprehensive consolidation architecture, unifying all database systems onto a single modern Oracle Exadata rack using virtualization. Each business system was deployed on a separate virtual machine (VM), ensuring:

• Resource Isolation: Separate IP ranges, administrative privileges, and security measures to comply with PCI-DSS standards.
• Hardware Optimization: Maximized utilization of the new Exadata platform’s capabilities.
• Scalable Flexibility: Easy addition of resources as needed.

During implementation, the client decided to increase the number of database systems by 50%, bringing the total to nearly 20 databases. This required the MPS team to recalculate capacity, reallocate resources, adjust schedules, and refine the migration plan ,while staying within the original budget and design constraints.

System Design

The system was designed with a modern architecture to ensure high availability, security, and optimal performance:

• At the Data Center (DC):
  • 1x Oracle Exadata Database X9M-2 EF Quarter Rack, including:
    • 2 Database Nodes for computation.
    • 3 Cell Nodes for Database storage.
• At the Disaster Recovery (DR) Site:
  • 1x Oracle Exadata Database X9M-2 EF Quarter Rack, including:
    • 2 Database Nodes for computation.
    • 3 Cell Nodes for Database storage.
• Cluster Configuration:
  • Both Exadata racks were configured as a cluster to ensure high availability and optimized database processing performance.
• Virtualization with RAC:
  • Each rack was divided into multiple Real Application Clusters (RAC) pairs, each deployed as virtualized instances distributed across the two physical Database Nodes to ensure high availability.
• Real-Time Synchronization:
  • The systems at the DC and DR sites were synchronized in real-time using Oracle Data Guard, ensuring security, compliance, and high availability.

Implementation Process

The implementation was divided into four main phases:

1. Survey and Assessment:
   • The MPS team conducted a detailed analysis of data volumes, storage capacity, and IOPS requirements for each system. Resource isolation, network ranges, and security permissions were clearly defined, with plans prepared for the 50% increase in database systems requested by the client.
2. Architecture Design:
   • A detailed VM allocation plan was created to avoid resource overlaps or network conflicts. The migration (cutover) plan was meticulously designed to minimize downtime, a critical factor for a bank operating 24/7. The design also accounted for future data growth.
3. Migration Testing:
   • The MPS technical team, including system engineers and Oracle database experts, set up the new Exadata system, configured VMs, allocated resources appropriately, and migrated data to corresponding VMs. Each migrated system underwent rigorous testing to ensure compatibility, performance, and error-free operation. This phase required close coordination with the business team to avoid unplanned disruptions. Benchmark tests were conducted to evaluate real-world performance, and the Exadata configuration was fine-tuned to ensure optimal performance, even with the additional 50% workload.
4. Cutover and Go-Live:
   • Detailed migration plans were developed for each system, reviewed, and approved by the client. Systems were migrated sequentially to the Exadata platform. The entire cutover and go-live process spanned several months to align with each system’s schedule.

Achieved Results

The project delivered impressive outcomes:

• Significant Performance Improvement: Overall database system performance increased by 30% to 50%, particularly for OLTP and large-scale data analytics tasks.
• Reduced Operating Costs: Operating costs decreased by 25% to 40% by eliminating legacy non-Oracle servers and centralizing management on a unified platform.
• Optimized Resource Utilization: Database system resource usage was reduced by 2 to 4 times compared to the original setup while maintaining full business performance.
• Security and Flexibility: The virtualization model ensured clear logical separation, meeting PCI-DSS standards, while allowing easy resource expansion.
• Consolidation Efficiency: A single Exadata X9M-2 Quarter Rack with 128 physical cores successfully hosted nearly 20 database systems, far exceeding initial expectations.

Key Lessons Learned

The database consolidation project was not only a technical endeavor but also a long-term strategic initiative. Key lessons include:

• Maintain Logical Separation: Preserving logical partitions while consolidating physical infrastructure optimizes performance and simplifies management.
• Plan for Scalability: Incorporating scalability into the initial design prevents reactive adjustments to client-driven changes.
• Rigorous Testing: Prioritizing post-implementation testing ensures system stability and readiness for complex business requirements.

Conclusion

Consolidating nearly 20 database systems onto a modern Oracle Exadata platform enabled the bank to modernize its data infrastructure, optimize costs, enhance processing speed, and standardize operations. With just an Exadata X9M-2 Quarter Rack and 128 physical cores, the bank successfully hosted nearly 20 database systems, demonstrating the platform’s superior capabilities. The latest Exadata X11M version offers up to four times the CPU cores of the implemented version, promising even greater processing power. This project is a testament to how a well-executed strategy can turn challenges into opportunities, empowering organizations to thrive in the digital era.

If your organization faces similar challenges, MPS Vietnam is ready to partner with you—from architecture consulting and planning to implementation and post-project optimization—ensuring your systems achieve maximum efficiency and return on investment.

References:

·         Oracle. (2021). Oracle Exadata Database Machine X9M-2 Data Sheet. https://www.oracle.com/a/otn/docs/exadata-x9m-2-ds.pdf

·         Oracle. (2021). Oracle Exadata Database Machine X11M Data Sheet. https://www.oracle.com/a/otn/docs/exadata-x9m-2-ds.pdf

·         Oracle.  Virtualized Exadata Database Machine. https://docs.oracle.com/cd/E73210_01/EMXIG/GUID-E98B6B35-2B66-42B8-8F2B-AD30EDF2B338.htm?#EMXIG327