Business Challenges

Lichu processes more than 23 million transactions every day and the cost of storage bursts due to rapid growth of transaction data.

The core payment business is processed in numerous sharding databases and tables in which data is frequently stored and split. This complicated O&M impacts business continuity and stability. The massive volume of payment data that has increased over time also increase the storage costs significantly.

Numerous MySQL instances are deployed to host different business modules, increasing the complexity of the O&M. In addition, the work load is unbalanced among business modules. Therefore, the computing resources of the instances cannot be fully utilized.

The capability of rapid elastic scaling is required for the database to deal with the explosion of new businesses and the highly concurrent traffic surges.

TP and AP businesses are hosted on different databases. It not only incurs additional costs but also introduces latency that leads to slow business analysis. The calculation of profit sharing takes too much time because statistical analysis involves databases of heterogeneous architectures.

A MySQL-compatible solution is required to support the business innovation with minimal application adaption costs.

Solution

The core payment module is deployed in a dedicated OceanBase cluster to ensure the stability of core business.

With read/write separation, the HTAP engine is capable of TP and AP at the same time. It ensures business stability and provides real-time data warehouse capabilities to improve the timeliness of operational analysis and decision-making.

Other business modules are deployed in a DBaaS OceanBase clusters to improve overall resource utilization based on multi-tenancy. OceanBase Migration Service (OMS) synchronizes payment orders to Kafka instances where the data is cleansed and written to the user center based on subscription.

As the native distributed database, OceanBase is deployed with a transparent and partitioned architecture. This helps avoid complicated sharding operations and guarantees data consistency because transactions are no longer processed across databases.

Multiple tenants are created to host business modules of different sizes to improve the efficiency of O&M.

Public cloud services are provided to support auto scaling all-in-one O&M and development management.

Compression algorithms are applied to significantly reduce storage costs.

Benefits

The storage space is reduced by 74% because of a high compression ratio.

Elastic horizontal scaling obtains the linear performance increasing.

The hybrid transaction/analytical processing (HTAP) engine provides 10 times better performance on AP workload.

Advanced and secondary compression algorithms are applied to reduce the storage costs by 75%, after the migration from MySQL to OceanBase Database.

The elastic public cloud can be quickly scaled out in response to the business growth and big promotion campaigns with ease.

The HTAP engine is introduced to support real-time data warehouse analysis. A lightweight real-time data warehouse that supports HTAP increases the efficiency of profit sharing calculation by 20 times.

The payment business is hosted in a dedicated cluster to ensure its stability, while other business modules are hosted in the DBaaS clusters. Resource utilization is increased based on multi-tenancy isolation.

As a result of partitioning, the database performance and stability are increased by 2 times.

Comprehensive ecosystems are provided to support system operation. 

• A unified console for centralized management is provided to improve the O&M efficiency.

• An end-to-end data migration tool is provided  to save customers' R&D costs for alternative data verification tools and achieve lossless cutover.