비용 최적화 및 지속 가능성 원칙

Cost Optimisation 필라 개요

# Cost Optimisation pillars:
# 1. Expenditure awareness  - visibility into what you spend
# 2. Cost-effective resources - right instance types, storage classes
# 3. Matching supply to demand - auto scaling, spot instances
# 4. Optimising over time - regularly review and adjust

# Example: undifferentiated heavy lifting
# Instead of managing your own Redis: use ElastiCache
# Instead of managing Kubernetes: use EKS or Fargate
# Managed services reduce operational overhead AND cost

리소스 적정 규모 조정

# Get Compute Optimizer recommendations for all EC2
aws compute-optimizer get-ec2-instance-recommendations \
  --filters Name=finding,Values=Overprovisioned

# Response includes:
# currentInstanceType: m5.4xlarge
# recommendedInstanceType: t3.large
# estimatedMonthlySavings: $280
# performanceRisk: VeryLow

# Also check EBS volumes:
aws compute-optimizer get-ebs-volume-recommendations \
  --filters Name=finding,Values=Overprovisioned

구매 모델 최적화

# Purchasing model comparison:
# On-Demand:       $0.192/hr (m5.large)   No commitment
# 1yr Reserved:    $0.114/hr              $0.78/hr effective
# 3yr Reserved:    $0.074/hr              Highest savings
# Compute SP:      ~$0.128/hr             Flexible family/region
# Spot:            $0.05-0.08/hr          Interruptible

# Savings Plans cover:
# - Compute Savings Plans: EC2 + Lambda + Fargate
# - EC2 Instance Savings Plans: specific family in one region

비용 최적화를 위한 Spot Instances

# ASG with mixed instances (On-Demand + Spot)
aws autoscaling create-auto-scaling-group \
  --auto-scaling-group-name my-mixed-asg \
  --mixed-instances-policy '{
    "LaunchTemplate": {"LaunchTemplateSpecification":{"LaunchTemplateId":"lt-12345","Version":"$Latest"},"Overrides":[{"InstanceType":"m5.large"},{"InstanceType":"m5a.large"},{"InstanceType":"m4.large"}]},
    "InstancesDistribution": {
      "OnDemandPercentageAboveBaseCapacity": 20,
      "SpotAllocationStrategy": "capacity-optimized"
    }
  }' \
  --min-size 2 --max-size 20 --desired-capacity 5

S3 스토리지 비용 최적화

# S3 lifecycle policy for cost optimisation
aws s3api put-bucket-lifecycle-configuration \
  --bucket my-data-bucket \
  --lifecycle-configuration '{
    "Rules": [{
      "ID": "auto-archive",
      "Status": "Enabled",
      "Transitions": [
        {"Days": 30, "StorageClass": "STANDARD_IA"},
        {"Days": 90, "StorageClass": "GLACIER_IR"},
        {"Days": 365, "StorageClass": "DEEP_ARCHIVE"}
      ]
    }]
  }'

태그 지정 및 비용 할당

# Enforce required tags with Config rule
aws configservice put-config-rule \
  --config-rule '{
    "ConfigRuleName": "required-tags",
    "Source": {"Owner":"AWS","SourceIdentifier":"REQUIRED_TAGS"},
    "InputParameters": "{\"tag1Key\":\"Project\",\"tag2Key\":\"Environment\",\"tag3Key\":\"Owner\"}"
  }'

# Query cost by tag in Cost Explorer
aws ce get-cost-and-usage \
  --time-period Start=2026-06-01,End=2026-06-30 \
  --granularity MONTHLY \
  --group-by Type=TAG,Key=Project

Sustainability Pillar 개요

# Sustainability improvement areas:
# 1. Right-size instances (reduce idle energy use)
# 2. Use Graviton (ARM) instances: 60% less energy than x86
# 3. Use Spot instances: uses otherwise idle capacity
# 4. Use managed services: AWS optimises their utilisation
# 5. Use serverless: no idle servers
# 6. Move to S3/EFS instead of EC2 instance storage
# 7. Implement data lifecycle (don't store forever)

Sustainability 및 비용을 위한 AWS Graviton

# Compare: m5.large (x86) vs m7g.large (Graviton3)
# m5.large:  $0.096/hr, 2 vCPU, 8 GB
# m7g.large: $0.0808/hr, 2 vCPU, 8 GB
# Savings: ~16% cheaper + 40% better performance

# Switch Lambda function to Graviton (arm64)
aws lambda update-function-configuration \
  --function-name my-function \
  --architectures arm64

# Lambda arm64 is 20% cheaper than x86
# Most Python, Node.js, Java functions work unchanged

유휴 리소스 제거

# EventBridge + SSM to stop dev instances nights/weekends
aws events put-rule \
  --name stop-dev-instances \
  --schedule-expression 'cron(0 22 ? * MON-FRI *)'

aws events put-targets \
  --rule stop-dev-instances \
  --targets '[{
    "Id": "StopDevInstances",
    "Arn": "arn:aws:ssm:us-east-1::automation-definition/AWS-StopEC2Instance",
    "RoleArn": "arn:aws:iam::123:role/EventBridgeRole",
    "Input": "{\"InstanceId\":[\"i-dev1\",\"i-dev2\"]}"
  }]'

Sustainability를 위한 데이터 수명 주기

# DynamoDB TTL for session data
# Add ttl attribute to items (Unix epoch timestamp)
aws dynamodb update-time-to-live \
  --table-name UserSessions \
  --time-to-live-specification Enabled=true,AttributeName=expiresAt

# Item will be deleted automatically after expiresAt timestamp
# Example: {'userId': 'u1', 'expiresAt': 1750000000}

# CloudWatch Logs: set 30-day retention
aws logs put-retention-policy \
  --log-group-name /aws/lambda/my-function \
  --retention-in-days 30

비용 최적화와 다른 Pillar의 비교

# Cost vs reliability trade-off example:
# Single-AZ RDS: $100/month, no HA
# Multi-AZ RDS:  $200/month, automated failover

# Decision: if database failure = $10,000/hour of revenue loss
# Even 1 event/year justifies Multi-AZ
# ($10,000 expected loss > $1,200/year extra cost)

# SAA-C03 exam approach:
# Meet the stated requirements FIRST
# Then choose the cheapest option that meets them

빠른 확인

강의 요약