Kubernetesの耐障害性を強化し、SLA違反を削減

The Solution

Results

• Over 90% reduction in CPU throttling and OOM issues — from multiple issues a day to one or two per month
• Drastically reduced mean-time-to-resolution (MTTR) for capacity-related issues
• Service owners receive precise, evidence-based recommendations to resolve issues quickly
• Identified significant cost savings in some clusters and reinvested them into under-resourced clusters
• Achieved a fully stable, resilient, and cost-effective environment with no budget impact
• Eliminated time-consuming manual right-sizing work across a dozen multi-regional EKS clusters

概要

課題

成果