Accelerating Digital Transformation Why Ping An Health Insurance is completely "de-oracle" database

With the increasingly fierce competition in the health insurance industry, the business model of simply providing protection services can no longer meet the diversified needs of the market. At the same time, how technological innovation can empower the insurance industry to accelerate its digital transformation has also attracted much attention.

"It is foreseeable that the core competitiveness of insurance companies in the future is not only the improvement of the efficiency of a single business, but the innovative development based on the integration of business model and technology driven by the company's strategy." Ma Rongqiang, CTO of Ping An Health Insurance, said in an interview with reporters on May 13 that insurance, as a low-frequency product, needs to ensure the accuracy of recommendations to improve the conversion rate, which requires insurance companies to have a series of capabilities such as building customer portraits, hierarchical management and labeling, so as to provide possibilities for business model changes. Based on this, abandoning the Oracle Database (Oracle Database, or Oracle, a relational database management system of Oracle) that has always been used in the industry in the past and reconstructing the core system of a new generation of business may be a new exploration and new paradigm to build a competitive moat.

The logic of building a new system

In Ma Rongqiang's view, due to the nature of the industry, the product innovation configuration link of the insurance business in the past was long and the product innovation cycle was long. Through the combing of a large number of business processes and business scenarios, the team formed several closed-loop domain business capabilities based on the architecture method of Domain-Driven Design, and built a new generation of core business system based on the originally independent business segments such as insurance protection and medical and health services.

Specifically, the "3+3+3" global architecture system built by Ping An Health Insurance refers to the "C-end + A-end + P-end (cooperation end, including B-end enterprises, third-party channels, ecological partners, etc.) of the front-end system" to reach customers and sales, the "individual insurance + group insurance + medical health" commercial general capability of the middle-end system, and the underlying support capability of "digital intelligence + business finance + technology application base" of the back-end system.

"Under the new system, according to specific business logic and needs, product developers can intelligently match products and services, combine them like building blocks, and quickly iterate and innovate products." Ma Rongqiang took the "super product factory" as an example to explain that there were 7 product factories on the original old system, and the products, services, and rights were all independent and fixed to each other. In the new system, all business types are abstractly integrated into an integrated product center platform, and the combination of insurance products and medical and health service rights is as simple as building blocks. The original product innovation cycle was 50 days, but now it can support the product to go online in 1 day at the earliest.

Talking about the reason for building a new system, Ma Rongqiang said frankly that although some people may question the necessity of building a new system, thinking that although the old system is stuck, it can be used reluctantly. But from bungalows to high-rise buildings, from bicycles to high-speed rail airplanes, these changes prove that the times are constantly progressing and technology is constantly evolving. In the same way, the construction of the new system is also to adapt to this rapidly changing era, and use new technologies to solve the problems that may arise now and in the future, so as to occupy a favorable position in the future competition.

On the basis of ensuring the smooth operation of existing businesses, it is not easy to reconstruct the entire system of core business. Xu Qing, deputy general manager of the customer channel service team of Ping An Health Insurance Technology Support Center, said that the first thing is that the existing insurance business has completed a closed loop, and the new architecture design goal requires insight into the essence of the business and the dismantling and reconstruction of the process. In the case of reshaping business logic and processes, how to complete the upgrade without affecting the existing experience and smooth operation of the business is a technical difficulty that needs to be overcome in the actual operation process.

"Another big challenge is that there are not many insurance companies in the industry that have completely removed Oracle from the core insurance system, and most of them will choose to move Oracle's stored processes to other databases, and it often takes 3 to 5 years, which also means that there is no more experience to learn from when reconstructing the new system." Xu Qing said.

The reporter learned that most insurance companies, including Ping An Health Insurance, used to use the centralized architecture of Oracle database. However, this model also has its obvious drawbacks, that is, the underlying data is complex, the coupling degree is high, and it is not easy to expand, which is not suitable for the current high-speed development of the business, and at the same time, the technical maintenance cost is high and difficult, and a lot of manpower and material resources need to be invested every year.

"The original centralized architecture model has been in use for more than 30 years, and it is clear that it cannot meet the needs of today's huge business volume. As a simple example, even if it's just to change a meme, we need to wait until midnight to lock the table with a centralized architecture. Wang Qi, general manager of the middle office service team of Ping An Health Insurance Technology Support Center, said that in the process of building a new system, the team chose the "Oracle" solution, deployed and landed a domestic database based on OceanBase, and adopted a microservice and distributed design architecture.

It is reported that the core business of medium and large health insurance companies is "de-Oracle" database, which is still a minority in the industry.

In this regard, Wang Qi said that there is also a transformation of "changing the soup but not the medicine" in the industry, that is, although the domestic database or 0 code database has been changed, the core architecture still relies on the logic in the database, and such a transformation has not brought substantial benefits. On the contrary, a small number of companies that choose a more point-like and partial module transformation method may have the greatest benefits in the long run, although it may be more difficult in the early stage.

"The distributed architecture reduces the operational costs of both database and IT by consolidating multiple database clusters into a single tenant database cluster. This transformation not only accelerates demand response, but also improves structuring and modularity, reduces coding effort, and reduces trial-and-error costs, resulting in both IT and business benefits, and our system stability is now 91.4% better. Wang Qi said.

What are the results of improving quality and efficiency

It is reported that Ping An Health Insurance will officially start the restructuring of its core business system at the end of March 2023, and complete the launch of the "good start" product demand at the end of November. At present, almost 100% of its individual underwriting and preservation services are running in the new system, and some of the traffic of group business has been switched to the new system.

Ma Rongqiang took the 2023 tax preferential project as an example, and said that it only took one and a half months from regulatory requirements to the actual product launch, which is undoubtedly a huge advantage for the industry compared to the half year usually required by the industry. This is largely due to the elasticity and pluggability of the architecture, which allows the team to respond quickly to requirements and leverage existing reusable components and modules to accelerate development.

For code that needs to be rewritten, Xu Qing said that the first attempt is to meet the needs by reusing and combining existing components, and only when new rules or special functions are really needed, will the code and configuration be used. This strategy not only improves development efficiency, but also reduces the likelihood of errors. At the same time, the AI architecture also follows the design concept of parallel, scalable, and unified scheduling, which means that no matter what scenarios and requirements are faced, the new system can quickly find and apply the most suitable model, and achieve efficient and accurate response through the elastic architecture.

It is reported that based on the accelerating digital transformation, Ping An Health Insurance has also launched the "Yixiang Easy Insurance" medical insurance product a few days ago, which is the industry's first medical insurance that actively protects the health of customers, focusing on digital health protection, and pre-managing customer health through dynamic health monitoring and early warning services.

When talking about the effectiveness of the new system on the company's operation, He Xinyi, head of the platform product group of the product design team of Ping An Health Insurance Technology Support Center, gave a set of data. Taking group insurance business as an example, the automatic entry rate has increased by 40%, and the processing time of the system for 100,000 people has increased by 500%; In terms of building the foundation for the integration of medical claims, the efficiency of automatic claims analysis has been increased by 140%. In terms of risk prevention and control capabilities, the proactive detection rate of production problems has increased to 63%, and at least half of the system problems have been prevented and improved based on active discovery, effectively reducing the customer complaint rate. The accuracy rate of the unified trust index of regulatory reporting reached 100%, and the error rate of Zhongbaodeng data dropped to 0.41%.