Project Background
The hospital operates multiple campuses and serves as a major medical institution in Luoyang. Under national digital-economy and localization strategies, it partnered with leading domestic vendors to build a healthcare localization application lab.
From infrastructure selection to production rollout, the hospital progressively introduced domestic technology into critical systems. By 2025, it had adapted 86 applications including HIS and EMR to the target ecosystem.
Challenges and Requirements
- All diagnosis and treatment systems run on a microservice-based smart-healthcare platform, so the database plan had to balance stability, performance, and resource efficiency.
- The migration covered 43 key systems, 55 databases, and 5 TB of structured data, while continuous downtime was not allowed to exceed 2 hours.
- After migration, the database platform still had to maintain or exceed the performance level of the former foreign stack in real clinical scenarios.
Implementation Approach
Six HA database clusters for core services
Highgo deployed six clusters using primary, synchronous standby, and asynchronous standby nodes with log-stream replication and automated monitoring, enabling around 10-second failover for single-node faults.
Phased migration with hot-cold separation
The migration used a full-load plus phased strategy, automation tools, multi-round validation, and hot-cold data separation to move 55 databases to domestic ARM-based Highgo clusters.
Read-write splitting and continuous tuning
After go-live, the project team optimized slow SQL, adjusted key parameters, applied storage compression policies, and routed read traffic to standby nodes to reduce pressure on the primary.
Project Highlights
- Completed localization of diagnosis-and-treatment core systems including HIS and EMR.
- Adopted a tailored hot-cold separation and phased migration strategy with bidirectional fallback paths and dual-environment protection.
- Used synchronous and asynchronous cluster combinations to balance high availability, performance, and cost, with around 10-second single-node failover.
- Implemented a deep read-write split design across business and database layers, reducing registration and pharmacy workload significantly.
- Supported around 8,000 patient visits per day during peak operation after go-live.
Customer Value
- Highgo reliably supports the hospital’s core systems, including daily handling of more than 8,000 outpatient and 3,000 inpatient visits through the HIS platform.
- Database load balancing improved patient-service experience by cutting average queue waiting time from 45 minutes to 28 minutes and surviving one-day concurrency of around 10,000 visits.
- The project proved that domestic databases can deliver high availability, stability, and performance in complex top-tier hospital scenarios.
- The program also received strong external recognition, including approval as a provincial healthcare localization innovation center and inclusion in national healthcare localization guidance.