Datadog Live with Devsisters Event Review

Devsisters has been operating its services using AWS since 2011, but existing tools like Zabbix and CloudWatch were complex to configure and maintain, and had difficulty responding to the variability of cloud environments. To solve this, they introduced Datadog, which provided practical help, such as quickly resolving a bottleneck that occurred during the launch of Cookie Run: OvenBreak.

Key use cases by team are as follows.

• Game Development Team: Monitors real-time metrics such as RPS and custom metrics like revenue/play statistics, and improves server performance through APM.
• Platform Team: Visualizes service dependencies in the MSA structure with Service Map, and manages external API status with Synthetic Monitoring and Service Check.
• DevOps Team: Monitors Kubernetes cluster events and resource usage, and manages service stability levels through SLO.

In terms of infrastructure management, they transitioned from a single cluster structure to operating individual clusters per game to prevent the propagation of failures. All infrastructure is managed as Terraform code, and clusters and add-ons are efficiently deployed and managed through their own custom-built EKS Terraform module.