cost-optimization

Skill: Cost Optimization

Expertise: Cloud cost analysis, right-sizing recommendations, reserved/spot instances, tagging strategy, cost alerting.

When to load

When investigating unexpectedly high cloud bills, right-sizing instances, or setting up cost attribution by team/environment.

Mandatory Cost Tags (every resource)

locals {
  cost_tags = {
    Project     = var.project      # e.g. "checkout"
    Environment = var.environment  # prod / staging / dev
    Team        = var.team         # e.g. "backend-team"
    CostCenter  = var.cost_center  # e.g. "eng-platform"
    ManagedBy   = "terraform"
  }
}
# Apply to every resource via merge(local.cost_tags, {...})

AWS Cost Discovery

# Top 10 most expensive services this month
aws ce get-cost-and-usage \
  --time-period Start=$(date +%Y-%m-01),End=$(date +%Y-%m-%d) \
  --granularity MONTHLY \
  --metrics BlendedCost \
  --group-by Type=DIMENSION,Key=SERVICE \
  --query 'ResultsByTime[0].Groups | sort_by(@, &Metrics.BlendedCost.Amount) | reverse(@) | [:10]' \
  --output table

# Cost by tag (Team)
aws ce get-cost-and-usage \
  --time-period Start=$(date +%Y-%m-01),End=$(date +%Y-%m-%d) \
  --granularity MONTHLY \
  --metrics BlendedCost \
  --group-by Type=TAG,Key=Team

# Idle EC2 instances (< 5% CPU, last 2 weeks)
aws cloudwatch get-metric-statistics \
  --namespace AWS/EC2 \
  --metric-name CPUUtilization \
  --dimensions Name=InstanceId,Value=<instance-id> \
  --start-time $(date -d '-14 days' --iso-8601) \
  --end-time $(date --iso-8601) \
  --period 1209600 --statistics Average

Right-Sizing Workflow

# Step 1: Get actual utilisation (Prometheus, last 7 days)
# CPU: avg and p99
avg(avg_over_time(instance:node_cpu_utilisation:rate5m[7d])) by (instance)
quantile_over_time(0.99, instance:node_cpu_utilisation:rate5m[7d])

# Memory: peak working set
max_over_time(node_memory_MemUsed_bytes[7d]) / node_memory_MemTotal_bytes

# Step 2: Recommendation formula
# New CPU = p99_cpu × 1.3 (30% headroom)
# New memory = peak_mem × 1.2 (20% headroom)

# Step 3: Find next smaller instance type
# AWS: use ec2-instance-selector
ec2-instance-selector --vcpus-min 2 --vcpus-max 4 \
  --memory-min 4 --memory-max 8 \
  --region eu-west-1 \
  --output table-wide