Elevate modern apps with TiDB.
One of the world’s top 50 property insurance companies, known for its comprehensive range of insurance products and services, is encountering significant challenges with their technology infrastructure. This company offers property damage, liability, credit and surety insurance, and operates through 36 subsidiaries and 337 branch offices. Their database is struggling to keep up with such extensive requirements.
The Challenge: Performance Bottlenecks with Large Data Volumes and Complex Logic
The policy system is the core of any insurance company, handling underwriting, policy issuance, endorsements, and more. It supports all internal and external channels. However, this insurance company’s policy system is facing significant challenges:
- Large Data Volume and High Load: The policy system is a logically centralized system featuring wide, large-core tables associated with every business. SQL queries can span thousands of lines.
- Performance Bottlenecks: Distributed sharding across data nodes can lead to performance bottlenecks. Migrating from mainframe to x86 hardware with centralized databases risks performance degradation.
- Complex Logic: The intricate logic of core operations frequently results in processing delays.
What are the Requirements?
As the infrastructure of a core system, the database must meet several key criteria to support large-scale data storage, high concurrency, and agile response times:
- Hybrid Transactional and Analytical Processing (HTAP): The system must handle both online transactions and real-time analytical processing.
- Scalable Storage: The new data system should support upwards of 100 terabytes (TBs), given current data volumes of 60TB+.
- High Throughput and Low Latency: It needs to sustain peak transaction throughputs above 500 transactions per second with peak response times under 10 milliseconds.
The Solution: Handling 64,000 QPS with an Average Response Time of 6 ms with TiDB
Since 2020, this insurance company has partnered with PingCAP to replace their existing Oracle database with TiDB, an open-source distributed SQL database. This transition covered 9 business systems, including the document management system, auto and non-auto insurance, quotation center, agricultural insurance claims, and more. By 2023, the migration of the core policy system was successfully completed and put into production. The new architecture looked something like this:
Since its launch, the new core policy system has operated stably, supporting the daily issuance of 250,000 policies and handling amounts exceeding $20 million. During peak business periods, TiDB’s performance has been outstanding, with peak queries reaching 64,000 per second and an average response time of just 6 milliseconds.
The Results: The Distributed Transformation of a Logically Centralized Policy System
Implementing the native distributed database TiDB validated the feasibility of distributing complex business logic. TiDB’s OLTP performance in handling large data volumes and intricate logic surpasses that of sharded databases and table-group-based distributed systems.
Ensuring Low Latency During High Concurrency
TiDB’s storage engine, which uses the LSM-Tree, is ideal for bulk inserts during peak periods. This solution eliminated Oracle’s previous write-wait issues and effectively addressed the write performance shortcomings of dual-node Oracle RAC under high concurrency scenarios. During peak periods, TiDB achieved over a 50% reduction in average response latency compared to Oracle databases.
Reducing Total Cost of Ownership (TCO)
The new system achieved a reduction in TCO across hardware, development, and operations, replacing mainframes with x86 reduced procurement costs. TiDB supports automatic sharding without the need for prior schema design, simplifying business development and cutting down on development costs. Additionally, TiDB’s visual operations management platform can manage hundreds of clusters simultaneously, reducing operational manpower costs.
Promoting Business Innovation
By transitioning to a modern, distributed SQL database like TiDB, the company not only solved existing performance bottlenecks but also laid a robust foundation for future business innovation. The seamless handling of massive data volumes and high throughput positions the company to explore new insurance products and services with confidence.
Enhancing System Reliability and Scalability
TiDB’s distributed architecture ensures that the system can easily scale to accommodate growing data volumes and increased transaction rates without compromising on reliability or performance. This scalability is critical for the company as it continues to expand its service offerings and reach.
Facilitating Real-Time Data Analytics
The HTAP capabilities of TiDB enable the insurance company to perform real-time data analytics alongside transactional processing. This dual capability ensures that decision-makers have access to up-to-date insights, facilitating more informed and timely business decisions.
Conclusion
This Top 50 Global Insurance Company’s move from Oracle to TiDB has been transformative. By overcoming challenges related to large data volumes, high concurrency, and complex logic, TiDB has ushered in an era of improved performance, reduced costs, and enhanced business capabilities. The successful implementation of TiDB highlights the potential of distributed databases in modern enterprise environments.
Through this collaboration, this company not only optimized its core policy system. It also set a benchmark for leveraging distributed SQL databases to drive digital transformation and business innovation. As organizations navigate the complexities of scalable, high-performance data systems, TiDB offers a compelling solution that combines reliability, scalability, and efficiency in a single powerful package.
By embracing TiDB, this Top 50 Global Insurance company has positioned itself at the forefront of technological advancement, ready to tackle the future with agility and confidence.
