Background

Haidilao decided to gradually migrate its core online business systems to the cloud in 2018. In the same year, the company opened its first smart restaurant in Beijing, marking the beginning of the all-round business intellgence driven digital transformation of its online and offline business. Today, Haidilao has established a well-functioning digital system and is taking steps towards refined digital management. However, new trends in the consumer market keep bringing about new challenges for its business digitalization. In response to the data explosion, which demands better database performance in terms of scalability, service availability, and response time, Haidilao has no choice but to upgrade its database infrastructure.

Challenges

1. Rising costs due to rapid business growth

To deal with traffic surges, Haidilao tried database and table sharding. However, the original architecture could not be scaled down automatically once the traffic returned to normal, resulting in a waste of resources. As the trend of digital driven transformation accelerated in the catering industry, the company needed to figure out how to cut costs without compromising the customer experience or affecting the performance of the operations and management data service. 

2. Service availability under surging business traffic

Based on MySQL databases with sharded architectures, the previous solution was complex and had stability issues. Systems such as membership and queue management systems were overloaded when  business traffic surged during peak periods. Haidilao had experienced database failures that caused service interruptions during peak hours.

3. Smart recommendations based on the real-time analysis performance

Haidilao provides smart recommendations based on the tastes and preferences of each customer. This relies on online analytical processing (OLAP) capabilities providing prompt recommendations. The original architecture, due to its complex business traces, suffered from synchronization latency and incurred additional storage costs.

Solution

1. Database consolidation via multitenancy and advanced compression technology

OceanBase Cloud uses the LSM-tree-based advanced compression technology to reduce data storage costs, utilizes multitenancy and hybrid deployment to make full use of system resources, and leverages hybrid transaction/analytical processing (HTAP) capabilities to effectively reduce data redundancy and save on the costs of data warehouse deployment.

2. Multi-level auto scaling

OceanBase Cloud uses a shared-nothing architecture with multiple replicas to ensure zero single point of failure (SPOF) and system continuity. Specifically, OceanBase Cloud supports smooth scaling at multiple levels. For example, during peak hours, Haidilao can scale the database at tenant and cluster levels. The multilevel auto-scaling feature eliminates the disfloorBanners of the sharding solution, such as insufficient scalability and poor reliability, ensuring the stability and service availability of Haidilao systems.

3. HTAP

With its HTAP capabilities, OceanBase Cloud can use one computing engine to simultaneously run online transaction processing (OLTP) and OLAP applications on multiple computing nodes in the system, achieving real-time analysis with minimal data synchronization latency. To process the membership-level data in the original database, Haidilao previously needed to load data from the database, cleanse data through the extract-transform-load (ETL) process, and send the data to a data warehouse where batch processing jobs were executed. After the calculation was completed, the result was synchronized back to the database. Now, Haidilao can use OceanBase Cloud to automate all these processes.

Benefits

1. Reduction in overall database costs

The overall costs of Haidilao's new database are reduced by 50%. While lowering costs, OceanBase Cloud optimally utilizes servers resources without compromising the service capabilities of the database.

2. Stable performance under traffic surges on holidays

The performance and stability of the database have significantly improved. The auto-scaling feature of OceanBase Cloud has enabled Haidilao's database to run smoothly, delivering stable service during peak periods such as the Chinese Valentine's Day promotion. The new database provides services 24/7, reaching a recovery point objective (RPO) of 0 seconds and a recovery time objective (RTO) of less than 30 seconds.

3. Improved computing power for real-time analysis

In addition to the simplified database architecture and cost reduction, OceanBase Cloud provides HTAP capabilities, which have improved the system's performance for real-time analysis by 45%. This capability allows Haidilao to provide smart recommendations by analyzing the tastes and preferences of each customer in real time.

The stable and reliable database infrastructure allows Haidilao to effectively complete its digital transformation, laying a solid foundation for future growth. The efficient management of the digital membership system brings significant benefits to Haidilao. The number of members has increased from 85 million in June 2021 to over 100 million today, with more than 80% of the total turnover contributed by members, indicating a healthy growth trend.