检测到您已登录华为云国际站账号,为了您更好的体验,建议您访问国际站服务网站 https://www.huaweicloud.com/intl/zh-cn
不再显示此消息
Weli Technology is a mobile Internet company focuses on time management mobile apps. It has couples of well-known calendar apps such as Chinese Calendar and Weli Kankan.
Weli applications have different instance parameters. Data for different applications is stored in multiple types of databases, so different migration solutions are needed.
Chinese Calendar databases and tables needed to be sharded manually. This wastes resources and risks of malicious code modification.
Weli Kankan has a lot of legacy data in log host. The rapid business growth of Weli Technology caused resource waste.
Three migration solutions are provided for applications with different requirements for the migration time. Multiple DRS data migration tests and drills are performed in advance to ensure smooth migration to Huawei Cloud.
Chinese Calendar uses three sharding methods: table sharding, database sharding, and table/database sharding. Based on the application requirements and its future 3-5 years growth plan, Huawei's database team analyzed the manual sharding logic of Chinese Calendar, collected statistics on the data volumes of database/table shards, selected a proper sharding key based on common query conditions, and synchronized nearly 1 billion data records to DDM and RDS for MySQL.
Huawei helped migrate 64 self-built MySQL instances of Weli Kankan to five Huawei RDS for MySQL instances based on thorough analysis of the monitoring metrics, read/write pressure, and data volume of the self-built instances.
Workloads were smoothly migrated to Huawei Cloud. Nearly 1 billion data records of Chinese Calendar have been synchronized without data loss or performance deterioration.
The 64 self-built MySQL instances of Weli Kankan were integrated into five Huawei RDS for MySQL instances. Data is evenly routed to balance read/write loads, reducing O&M costs and pressure.
Data in Chinese Calendar is migrated to DDM and RDS for MySQL through database/table sharding, simplifying code and improving read/write performance.