국내기사
Autonomous PostgreSQL operator for self-healing and scaling in multi-domain Kubernetes environments = 다중 영역 Kubernetes 환경에서 자체 복구 및 확장을 위한 자율형 PostgreSQL 오퍼레이터
The increasing adoption of Kubernetes for cloud-native applications has intensified the need for reliable and autonomous management of stateful database services, particularly in multi-zone environments where failures and topology changes are expected. Although Kubernetes Operators offer declarative automation for managing complex workloads, their effectiveness in handling production-grade database lifecycles remains insufficiently explored. This paper presents an experimental evaluation of an autonomous PostgreSQL Operator deployed in a multi-zone Kubernetes cluster, focusing on its ability to manage key operational scenarios without manual intervention. The study investigates horizontal scaling, self-healing and automated failover, configuration-triggered zone relocation, and rolling upgrades through controlled configuration updates and fault injection. Operator behavior is analyzed using Kubernetes events, pod lifecycle transitions, and cluster state observations. Results show that scaling operations are executed in a consistency-preserving manner through synchronized replica initialization, primary failures are resolved autonomously via safe replica promotion, topology changes correctly trigger zone-aware re-orchestration, and rolling upgrades are performed sequentially while maintaining service availability. Across all scenarios, the Operator relies solely on Kubernetes-native declarative control loops to reconcile observed and desired system state. These findings demonstrate that PostgreSQL Operators can function as effective autonomous control systems for managing stateful databases in multi-zone Kubernetes environments, reducing operational complexity and supporting resilient, production-grade cloud-native deployments.