Cut Costs 25% and Boost Performance 70%: Retail Giant’s OceanBase Migration

Background

WanJia Shuke (万家数科商业数据有限公司) is the information technology subsidiary of China Resources Vanguard, focusing on retail. It provides data‑as‑a‑service, SaaS products, business insight consulting, and precise marketing solutions for retailers and their ecosystems.

Technical Challenges

Challenge 1: Parallel Systems

Multiple legacy systems (IBM Informix, Oracle, MySQL, etc.) coexist, creating data silos, requiring custom adapters and middleware, and increasing integration effort and resource consumption.

Challenge 2: Degraded User Experience

Both internal and external users demand personalized features, faster response times under high concurrency, and easier usability, which the existing architecture struggles to provide.

Challenge 3: Complex Linkage

Complicated data flows make fault isolation difficult, leading to performance bottlenecks and heightened security risk.

Challenge 4: Limited Scalability

The original databases cannot meet growing performance demands; scaling the system incurs high hardware and labor costs.

Strategy and Measures

The technical team addressed the issues from hardware, operating system, database, and development philosophy perspectives, selecting domestic X86/ARM hardware, OSes such as Loongson and Kylin, and adopting Domain‑Driven Design. After extensive evaluation, they replaced the legacy databases with the native distributed database OceanBase.

1. Introducing OceanBase and Benefits

Cost saving: high‑compression storage reduced capacity by ~60%, hardware cost cut by 50%, overall business cost down ~25%.

Resource utilization: multi‑tenant clusters improve resource isolation and reduce fragmentation.

Improved resilience and development efficiency: unified tech stack lowers development difficulty and speeds delivery.

Performance boost: system throughput increased by 70% and real‑time reporting became feasible.

Operations efficiency: platform‑based DB management enables DBA‑free operations, reducing tool development and maintenance effort.

2. Database Migration Experience

The team used OMS (OceanBase Migration Service) to migrate from a MySQL sharding cluster to OceanBase. Migration followed a read‑write split strategy: first migrate read traffic, then write traffic, ensuring high stability and a seamless user experience.

Key steps included:

Adapting applications for dual data sources.

Batch‑wise migration with each step completing within 10 seconds, minimizing impact.

Handling schema changes such as merging tables, verifying unique keys, and redefining primary keys (replacing Snowflake‑based keys with auto‑increment ranges).

During migration, the team addressed challenges like unique‑key conflicts, large‑table partitioning, and hotspot SQL optimization.

3. Integration with Flink Ecosystem

To achieve real‑time data processing, the team integrated Flink with OceanBase via Debezium‑format change data capture. OMS provided a visual control panel, supporting point‑in‑time synchronization and low‑maintenance operations. Flink streams processed data in under 2 seconds, delivering timely updates to downstream systems.

4. Optimization Cases

Using OceanBase’s OCP and ODC tools, the team diagnosed high‑frequency SQL latency, identified RPC‑heavy execution plans, and resolved them by creating table groups to eliminate RPC calls.

Issues Encountered During Operation

Scenario

Business logic used INSERT INTO ... ON DUPLICATE KEY UPDATE to avoid constraint conflicts.

Problem Points

MySQL logged the operation as UPDATE, while OceanBase logged it as DELETE + INSERT, causing downstream consumers to misinterpret the data flow and potentially produce errors.

Conclusion

By migrating to OceanBase, WanJia Shuke achieved significant cost reductions, improved resource utilization, enhanced system resilience, and delivered up to 70% performance gains. The migration also streamlined operations through platform‑based DB management and enabled real‑time analytics via Flink integration.