Gome Group’s Cloud Computing and Operations Automation Practices

Author Introduction: Si Yu, with experience at Alibaba, Xiaomi, and Ping An, now leads cloud computing and operations automation at Gome Group, aiming to improve operational efficiency and reduce costs through technology.

Gome Overview: Founded in 1987, Gome is China’s largest home‑appliance retailer with 1,500 stores and 300,000 employees, now expanding into real‑estate, fintech, smartphones, smart manufacturing, internet, and offline retail, requiring extensive technical support.

New Operations Thinking: Gome focuses on reducing operational costs and increasing efficiency by consolidating resources into a unified "Gome Cloud" IAAS platform, centralizing development teams, and providing shared services to all subsidiary businesses.

Gome Cloud Practice: Established in June 2016, Gome Cloud offers IAAS services internally and leverages third‑party clouds for rapid deployment, enabling on‑demand resource provisioning, unified procurement, and a shift from cost center to profit center.

IAAS Product Architecture: The platform is divided into three layers – User Layer (SRE portals and API), Control Layer (resource scheduling, billing, and management), and Service Layer (CMDB, GMSTACK, RDS, SLB, VPC, CDN, VPN, Monitor, MIX‑Cloud).

Control & Command Layer: Implemented in Go, the Command service orchestrates API calls, supports synchronous, asynchronous, and concurrent requests, and integrates HTTP, HTTPS, and RPC via the Thrift framework.

Monitoring: Initially using Zabbix for thousands of hosts, Gome later adopted Open‑Falcon (a Go‑based system) for high‑performance monitoring, including alarm convergence and direct VM metrics collection.

Automation Standards & CI/CD: Gome standardizes operations, builds an automation ecosystem around CI/CD, and provides CMDB, asset management, and configuration services that feed into deployment pipelines.

Deployment Practices: Uses LVS/Nginx based load balancing, supports HTTP gray‑release by gradually shifting traffic, and integrates Docker for containerized workloads, aiming for 50% Docker adoption.

CMDB as SaaS: Provides API‑driven asset and configuration data, deployed both internally and as an external SaaS offering for other subsidiaries.

Gray Release Example: A ten‑node cluster can release to two nodes, remove them from the L7 load balancer, wait for connections to drain, then roll out updates, minimizing user impact.

Overall, Gome’s cloud and automation initiatives illustrate how a traditional retailer can modernize its IT infrastructure, achieve cost savings, improve service delivery, and foster a DevOps culture across a large, diversified enterprise.