Automated Database Operations Platform Overview

This article introduces an automated database operations platform designed to shift routine DBA tasks to self‑service for developers and fully automate DBA‑related processes.

The motivation stems from the heavy workload of DBAs handling work orders, large‑table DDL, cluster provisioning, scaling, and data migration. Repetitive operations reduce efficiency, so the team analyzed requirements and decided to separate data‑related tasks (self‑service) from DBA‑related tasks (automation).

The new architecture mirrors production environments: a front‑end SLB for load balancing, an RDS core composed of the RDS engine, dynamic configuration center Lion, migration tool Puma, and the Zebra JDBC‑based connection pool. The database layer uses an MHA+MySQL setup, and a lightweight Job Center executes system‑level commands.

To avoid maintaining separate legacy systems, all components (servers, middleware, databases) are hosted on the same online infrastructure, with DBA responsibilities delegated to the platform and server management handed to business operations.

A stateless workflow center manages processes and tasks, ensuring idempotent sub‑tasks and using MySQL‑based lightweight locks for distributed execution. The Job Center pulls tasks from a MySQL queue, forks processes, and handles callbacks.

The platform supports one‑click cluster provisioning, scaling, backup/restore, traffic control, dynamic migration/sharding, and single‑table splitting. Dynamic migration follows four steps: seed data transfer, incremental data transfer, account/permission migration, and data source switch. Incremental migration can use a MySQL middle‑machine or Puma, each with trade‑offs.

During data source switching, the system performs permission verification, table locking within a transaction, checksum‑based consistency checks, source switch, killing blocked queries, and resetting slaves to ensure safety.

Single‑table sharding is achieved by configuring sharding rules, dumping data, setting up incremental sync, enabling dual‑write, and disabling the old table.

Developers benefit from a SqlEditor module that integrates query, edit, permission, and data masking functions, offering unified system‑level access accounts, centralized permission requests, enforced access policies, and enhanced audit and security controls.