Scenarios

Transparent sharding for transactions and payments

Transactions and payments are a core business for Ant Group. Initially, Ant Group used a sharding solution that provided horizontal scaling and grayscale upgrade capabilities for its core business, which greatly reduced systemic risks. However, as the Alipay business grew rapidly, the capacity of a single machine became insufficient. In a sharding architecture, this problem could be solved only by further splitting the data into M × N parts. This process was time-consuming, labor-intensive, and involved high technical risks.

The OceanBase database solves this problem using partitioned tables and their horizontal scaling capabilities. You can discretize each of the original N data portions into M data partitions using partitioned tables. This breaks the capacity limit of a single machine. The splitting process is completed at the data layer and does not require application modifications, which significantly reduces costs and technical risks. Additionally, the partitioned table solution in OceanBase allows users to ignore data distribution details. It supports distributed transactions across partitions and joins between multiple partitions. This provides transparent access to the distributed database.

Multi-dimensional queries for membership systems

Many enterprises have membership systems that store important user information. This information must be accessible from multiple dimensions. Examples include querying user details by user ID, querying user details by mobile phone number, or grouping users by age. This type of multi-dimensional query, especially a query on a non-partition key column, is a common problem in distributed scenarios with massive data volumes. This can typically be solved by creating shadow tables on non-partition key columns. However, this solution is a compromise at the application layer and does not use the native capabilities of the database. The OceanBase database can use its partitioned table feature to discretize and distribute data across multiple nodes in a cluster, which satisfies some query requirements. It can then use the strong consistency global index feature to meet other dimensional query and analysis needs. This implements data partitioning and horizontal scaling at the database layer and is transparent to the user.

Batch processing systems

Batch processing systems in many industries often involve numerous batch operations. These include complex calculations that join multiple large tables and involve numerous data updates. In batch processing, the data volume for each operation is large, many large tables must be joined, complex queries are frequent, and the volume of data updates is high. This causes single-point performance bottlenecks in traditional centralized databases. The cost of vertical scaling is prohibitively high. The OceanBase database SQL engine, developed over nearly 10 years, has excellent SQL execution and distributed computing capabilities. It can support complex hybrid transactional and analytical processing (HTAP) applications. Because OceanBase is a native distributed database, it can solve the single-point performance bottleneck problem of traditional databases. It can also reduce the costs associated with scaling out.