vivo's Database Operations Platform: Challenges and Solutions in the Cloud-Native Era

This article presents vivo's database operations platform (DaaS) and its response to three major challenges in the cloud‑native era: managing massive numbers of database instances, achieving elastic resource scaling, and protecting personal privacy data.

To handle large‑scale instance management, vivo built the DaaS platform emphasizing self‑service work orders and self‑healing capabilities, which reduced manual O&M workload by 92% and enabled automatic analysis or handling of 70% of database alerts.

For resource elasticity, vivo adopted a mixed‑deployment strategy (multiple instances, versions, and packages per physical machine), developed a multi‑threaded Redis Cluster scaling tool to accelerate and stabilize expansion/contraction, and implemented soft isolation via configuration and capacity prediction to address resource contention.

Cost optimization is further achieved through automatic package tuning: the platform scans for shrink‑eligible instances, sends shrink tickets for manager approval, and executes the shrink, saving over a hundred terabytes of space in four months.

Regarding privacy protection, vivo provides a full‑chain solution: marking privacy fields during development, periodic scanning and manual verification of existing tables, and automatic encryption of query results and exports. For MySQL, a transparent encryption scheme using gh‑ost for historical data conversion and a MySQL proxy enables zero‑downtime, zero‑code‑change encryption of live data.

The article concludes with a three‑stage outlook for fault handling (expert‑based → AI‑assisted → AI‑led), resource management (mixed instances → container‑based → storage‑compute separation), and privacy governance (unstructured‑data identification and accuracy improvement via human‑AI labeling).