How Nexus Financial Services Reduced Transaction Processing Time by 75% Using Cloud-Native Architecture

Overview

Nexus Financial Services, headquartered in Singapore, operates as a mid-sized payment processor serving e-commerce platforms, retail chains, and fintech startups across Southeast Asia. Founded in 2008, the company had grown to process over 2 million transactions daily, with peak volumes reaching 15,000 transactions per second during major sales events like Singles' Day and Black Friday.

The company's infrastructure, built in 2011 on a traditional three-tier architecture with Oracle databases, IBM WebSphere application servers, and IBM AIX powered servers, had served well for over a decade. However, by 2024, the limitations had become untenable. Transaction processing times had increased from 200 milliseconds in 2018 to over 1.2 seconds by mid-2024. Customer complaints were rising, and two major clients had threatened to move their business to competitors.

This case study examines how Nexus Financial Services transformed their entire technology stack over 18 months, achieving performance metrics that exceeded their original targets while maintaining 99.99% uptime throughout the migration.

Challenge

The challenges facing Nexus Financial Services in 2024 were multifaceted and interconnected. The technical debt accumulated over 15 years of incremental additions had created a system that was increasingly difficult to maintain, scale, and enhance.

Performance Degradation

The most immediate challenge was performance. The monolithic architecture, originally designed for thousands of transactions per day, was struggling under the weight of millions. Database queries that once took milliseconds now took seconds. The application's single-threaded transaction processing meant that during peak periods, transactions queued up behind slower operations, creating cascading delays.

Internal metrics from late 2024 showed that average transaction processing time had reached 1.2 seconds—with individual transactions occasionally taking over 8 seconds during peak load. More concerning was the inconsistent user experience: the same transaction might process in 400 milliseconds or 4 seconds depending on overall system load, making it impossible to guarantee Service Level Agreements to clients.

Scale Limitations

The second major challenge was inability to scale. Nexus's infrastructure was provisioned for 2x their then-current volume, but the architecture itself had hard limits. The underlying Oracle database could only scale vertically, and adding more powerful hardware was reaching diminishing returns—and significant cost.

When projected growth was factored in, Nexus estimated they would need to handle 5 million transactions daily within 3 years. Their current infrastructure couldn't achieve this without complete replacement. Vertical scaling to the necessary level would have cost over $2 million annually in hardware and support contracts alone—a cost that was neither sustainable nor strategic.

Operational Burden

The third challenge was operational. Deploying changes to the production environment required 6 engineers working over a weekend, coordinating manual database updates, application deployments, and configuration changes. A typical deployment carried a 15% risk of introducing issues that required rollback.

The system lacked automated testing, CI/CD pipelines, and infrastructure-as-code practices. Each environment—development, staging, production—was configured manually by senior engineers, making consistency impossible to guarantee and debugging increasingly difficult.

Goals

Nexus Financial Services established clear, measurable goals for their transformation initiative, internallycodenamed Project Velocity. These goals were approved by the board and formed the basis for measuring success.

The primary goal was to reduce average transaction processing time from 1.2 seconds to under 300 milliseconds—a 75% improvement. This target was chosen based on competitive benchmarks and client requirements. Secondary goals included achieving these performance improvements while maintaining 99.99% uptime, reducing infrastructure costs by 30%, and enabling deployment of new features in hours rather than days.

Beyond technical goals, Project Velocity had strategic objectives: reducing time-to-market for new payment features from 3 months to 2 weeks, enabling automatic scaling during traffic peaks, and building a foundation for future technologies including real-time fraud detection using machine learning.

Approach

Nexus Financial Services' approach to transformation combined industry best practices with careful consideration of their specific constraints—existing client relationships, regulatory requirements, and organizational capabilities.

Architecture Decision

After evaluating multiple approaches including refactoring to serverless and multi-cloud strategies, Nexus chose a Kubernetes-based microservices architecture on AWS. This decision was based on several factors: existing AWS expertise within the team, the managed nature of Amazon EKS reducing operational burden, and the ability to leverage AWS's financial services competency certifications for regulatory compliance.

The architecture comprised 15 independent microservices, each responsible for a bounded context within the payment processing lifecycle. These included services for transaction ingestion, fraud scoring, ledger management, settlement processing, notification handling, and client management. Each service maintained its own database schema, with asynchronous event-driven communication for cross-service operations.

Migration Strategy

Nexus adopted a strangler fig pattern to enable incremental migration while maintaining business continuity. Rather than a parallel Big Bang migration, services were migrated one at a time, starting with the lowest-risk components and building toward the most critical transaction processing core.

This approach allowed the team to validate each migration before proceeding, learn from production traffic patterns, and build confidence in the new architecture—all while clients continued to use the system without disruption. The migration was planned across five phases spanning 18 months.

Team Organization

Project Velocity required organizational change alongside technical transformation. Nexus restructured from a traditional hierarchical team into four cross-functional squads, each responsible for specific microservices and empowered to make architectural decisions within their domain.

Each squad included backend engineers, database engineers, DevOps specialists, and QA engineers. This structure enabled autonomous decision-making while maintaining shared standards through guilds and technical leadership groups.

Implementation

The implementation of Project Velocity spanned 18 months, from initial planning in October 2024 to full migration completion in March 2026. Here's how each phase unfolded.

Phase 1: Foundation (October 2024 – January 2025)

The first phase focused on establishing the foundation for the new architecture. The team set up Amazon EKS clusters across three availability zones, implemented service mesh using AWS App Mesh, and established CI/CD pipelines using GitHub Actions and ArgoCD.

Infrastructure-as-code using Terraform ensured consistent environments across development, staging, and production. The team also implemented comprehensive logging using the ELK stack and distributed tracing using AWS X-Ray—capabilities that would prove invaluable during problem-solving in later phases.

Perhaps most importantly, Phase 1 established the patterns and practices that would govern all subsequent development: automated testing at every level, code review requirements, deployment checklists, and on-call practices.

Phase 2: peripheral Services (February – May 2025)

The second phase migrated peripheral services including notification handling, client onboarding, and analytics reporting. These services were lower-risk—issues would impact functionality but not core transaction processing.

The migration of the notification service was particularly instructive. The team discovered that the legacy system was sending duplicate notifications to approximately 3% of transactions—a bug that had existed for years but was never identified. Fixing this bug reduced notification-related client support tickets by 40%.

The analytics service migration demonstrated the power of the new architecture. Reports that previously took 45 minutes to generate now completed in under 2 minutes, enabling real-time business intelligence that was previously impossible.

Phase 3: Core Transaction Processing (June – October 2025)

The third phase tackled the core transaction processing service—the heart of Nexus's business. This required careful design of the new service, extensive load testing using Gatling with synthesized traffic patterns, and implementation of circuit breakers and graceful degradation patterns.

The team implemented a dual-write pattern, writing to both the legacy and new systems during a transition period. This enabled comparison of results and rollback if issues were detected. For three months, every transaction was processed twice—once on each system—with automated reconciliation checking for discrepancies.

When the core service was cut over to the new architecture on October 15, 2025, the team was prepared for issues. Despite this preparation, the first hour revealed unexpected latency patterns that required three iterations to resolve. By hour 24, performance had stabilized—and exceeded targets.

Phase 4: Ledger and Settlement (November 2025 – January 2026)

The fourth phase migrated the ledger and settlement services—the systems of record for financial transactions. These were the most critical and required the highest confidence in the new architecture.

Given the financial implications, the team implemented additional safeguards: extended dual-write periods, enhanced reconciliation checks, and manual audit processes during the transition. The migration was scheduled over the New Year period when transaction volumes were lowest.

The ledger migration revealed a subtle but important difference in the new system's handling of decimal precision. The legacy system stored amounts with 4 decimal places; the new system used 2 decimal places internally but maintained 4 in the database. This difference, caught during reconciliation testing, could have caused cumulative discrepancies in settlement amounts over time.

Phase 5: Optimization and Legacy Sunset (February – March 2026)

The final phase focused on optimization and shutting down the legacy infrastructure. The team conducted extensive performance tuning: right-sizing Kubernetes node pools, optimizing database queries, implementing caching where appropriate, and tuning garbage collection parameters.

Post-migration analysis showed several opportunities for optimization that had been missed during initial implementation. Database connection pooling was adjusted, reducing database CPU utilization by 30%. Caching was implemented for frequently-accessed reference data, reducing database queries by an additional 25%.

The legacy infrastructure was decommissioned on March 15, 2026, exactly 18 months after Project Velocity began. The final decommissioning was anticlimactic—a server was shut down, and nothing happened. That nothingness was the point: the transition was complete.

Results

The results of Project Velocity exceeded Nexus Financial Services' original targets across every metric. The transformation delivered not just technical improvements but business outcomes that positioned the company for their next chapter.

Metrics

The metrics tell a compelling story:

• Transaction Processing Time: Reduced from 1.2 seconds to 280 milliseconds—an improvement of 77%, exceeding the 75% target.
• Infrastructure Costs: Reduced by 40%, from $180,000 monthly to $108,000—while handling 50% more volume.
• Deployment Frequency: Increased from bi-weekly to multiple times daily, with average deployment time reduced from 8 hours to 15 minutes.
• Uptime: Maintained 99.99% uptime throughout migration, with zero unplanned downtime in the 12 months since full migration.
• Time-to-Market: Reduced from 3 months to 10 days for new payment features.
• Peak Capacity: Successfully handled 22,000 transactions per second during 2025 Singles' Day—a 47% increase over previous peak.

Beyond these quantitative metrics, qualitative improvements were equally significant. Engineering team satisfaction scores increased from 3.2 to 4.4 out of 5. Client satisfaction scores improved, with two of the three clients who had threatened to leave instead expanding their contracts.

Lessons

The transformation of Nexus Financial Services offers several lessons for organizations embarking on similar journeys.

Start with Why

Understanding the business purpose of technical transformation is essential. Project Velocity succeeded because the entire organization understood not just what was changing, but why. Every team member could articulate how their work contributed to client outcomes and business results.

Incremental Migration Works

The strangler fig pattern of incremental migration proved invaluable. By migrating service-by-service, the team could validate each step, learn from production traffic, and build confidence progressively. The alternative—parallel development and Big Bang cutover—would have increased risk and reduced learning.

Invest in Observability

The investment in comprehensive logging, metrics, and distributed tracing paid dividends throughout the migration. When issues arose—and they did—the team's ability to quickly identify the root cause prevented minor issues from becoming major incidents.

People Matter Most

The technical architecture is only as good as the team operating it. Nexus's investment in training, certification, and building cross-functional teams created an organization capable of operating and continuously improving the new system.

Document Decisions

Architecture Decision Records (ADRs) documenting each significant decision proved valuable. When later issues arose or new team members joined, these records provided context for understanding the system—not just what was decided, but alternatives considered and why the chosen approach was selected.

The transformation of Nexus Financial Services demonstrates that even deeply entrenched legacy systems can be successfully modernized without business disruption. The keys are careful planning, incremental execution, organizational change aligned with technical change, and relentless focus on outcomes that matter to customers and the business.