1. Overview (Project Background)

SmartScore is South Korea's largest golf platform with over 2 million users. Given the service characteristics with concentrated real-time traffic for reservations, rounds, settlements, and notifications, stability and predictable cost structure were critically important.

In the existing AWS environment, costs for Compute, RDS, NAT Gateway, and Confluent SaaS continued to rise, increasing operational burden.

Accordingly, SmartScore decided to migrate the entire system to Oracle Cloud Infrastructure (OCI) for performance optimization and cost reduction. The migration project was completed in 3 months using a minimal downtime approach to minimize service impact.

2. Challenge (Problem Definition)

• AWS cost surge: Increased RDS IO, storage, backup costs, NAT Gateway, and Data Transfer expenses

• API processing delays due to limitations in handling peak-time reservation traffic

• Inefficient Confluent SaaS cost structure: Annual costs surge with increased message volume

• Lack of FinOps visibility making cost forecasting and management difficult

• High infrastructure cost structure limiting new service expansion

3. Solution (Resolution Approach)

• Performed full system migration using minimal downtime approach within 3 months (Full Lift-and-Optimize).

• Compute & API Server Migration: Stabilization based on E4/E5 Compute and introduction of OKE auto-scaling architecture.

• Database Optimization: Cost reduction in IOPS and storage through OCI MySQL and Block Volume tier separation.

• APAC's First OCI Streaming with Apache Kafka Implementation: Complete replacement of existing Confluent SaaS.

• Integrated Logging & Monitoring: Cost reduction compared to CloudWatch and enhanced failure detection and response speed.

• FinOps Framework Based on Universal Credit: Cost fixation and improved cost visibility through service-level tagging.

4. Result (Achievements)

• Successfully migrated entire system using minimal downtime approach within 3 months.

• 30-45% reduction in overall cloud costs compared to AWS.

• Up to 50% cost reduction in database area.

• Secured structure for annual savings of hundreds of millions of won through Confluent SaaS elimination.

• APAC's first OCI Streaming (Kafka) implementation with 25% reduction in message processing delay rate.

• 30% improvement in API response delay rate during reservation peak times.

• 35% reduction in operational workload.

• Secured cost predictability enabling stable service expansion.