Cloud Cost Optimization: FinOps Best Practices ile %40 Tasarruf

Cloud Cost Optimization: FinOps Best Practices ile %40 Tasarruf

Cloud bill every month artıyor. "We migrated to cloud to save money" demiştiniz. Şimdi on-premise'den pahalı. FinOps ile cost control geri kazanın.

FinOps Nedir?

Financial Operations - Cloud financial management. Engineering, Finance, Business collaboration. "Unit economics" odaklı.

Common Cost Culprits

1. Idle Resources

Dev environment 7/24 açık. Gece, weekend kullanılmıyor. %65 waste.

Solution:

# Lambda function - Auto stop dev instances
import boto3
ec2 = boto3.client('ec2')

def lambda_handler(event, context):
    # Stop dev instances at 6 PM
    instances = ec2.describe_instances(
        Filters=[{'Name': 'tag:Environment', 'Values': ['dev']}]
    )
    instance_ids = [i['InstanceId'] for r in instances['Reservations'] 
                    for i in r['Instances']]
    if instance_ids:
        ec2.stop_instances(InstanceIds=instance_ids)
    return {'statusCode': 200}

Savings: $5,000/month → $1,500/month (%70 azalma)

2. Oversized Instances

t3.2xlarge instance, avg CPU %15. Right-size to t3.large → %50 cheaper.

Tools: AWS Compute Optimizer, Azure Advisor

3. Unattached EBS Volumes

Instance sildin, volume kaldı. $0.10/GB/month. 10TB = $1,000/month waste.

# Find unattached volumes
aws ec2 describe-volumes \
  --filters Name=status,Values=available \
  --query 'Volumes[].{ID:VolumeId,Size:Size,Type:VolumeType}'

4. Data Transfer (Egress) Fees

AWS data transfer out: $0.09/GB. 100TB/month = $9,000!

Solutions:

• CloudFront CDN kullan (cheaper egress)
• Same-region architecture (inter-AZ free)
• S3 Transfer Acceleration avoid (pahalı)

Reserved Instances & Savings Plans

On-Demand vs Reserved:

• t3.medium on-demand: $0.0416/hour = $30/month
• t3.medium reserved (1 year): $0.0270/hour = $19.50/month
• Savings: %35

Strategy:

1. Baseline workload → Reserved instances
2. Variable workload → Spot instances (%90 discount!)
3. Burst → On-demand

Spot Instances

AWS Spot = unused capacity, %70-90 discount. Stateless workloads için perfect.

# Kubernetes cluster autoscaler with spot
apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-autoscaler-priority-expander
data:
  priorities: |-
    10:
      - .*-spot-.*
    1:
      - .*-ondemand-.*

Spot önce scale, fallback on-demand.

FinOps Metrics

1. Unit Economics

Cost per user, cost per transaction, cost per request.

Example: $50,000 cloud bill / 1M users = $0.05 per user

2. Cloud Efficiency Score

(Utilized resources / Provisioned resources) * 100

Target: >75%

3. Waste Percentage

Idle resources, unattached volumes, old snapshots.

Target: <10%

Tools

• AWS Cost Explorer: Built-in cost analysis
• CloudHealth: Multi-cloud cost management
• Kubecost: Kubernetes cost visibility
• Infracost: Terraform cost estimation
• Vantage: Cost reporting & alerts

Tagging Strategy

Resource tagging ile cost attribution:

tags = {
  Environment = "production"
  Team        = "backend"
  Project     = "api-v2"
  Owner       = "[email protected]"
  CostCenter  = "engineering"
}

Team bazında cost visibility. Showback/Chargeback.

Budget Alerts

# AWS Budget
aws budgets create-budget \
  --account-id 123456789 \
  --budget file://budget.json \
  --notifications-with-subscribers file://notifications.json

# budget.json
{
  "BudgetName": "Monthly-Engineering-Budget",
  "BudgetLimit": {
    "Amount": "10000",
    "Unit": "USD"
  },
  "TimeUnit": "MONTHLY"
}

Optimization Checklist

✅ Idle resources kapatıldı
✅ Right-sizing yapıldı
✅ Reserved instances/Savings Plans
✅ Spot instances (uygun workloads)
✅ Old snapshots/backups temizlendi
✅ Data transfer optimize edildi
✅ Resource tagging %100
✅ Budget alerts configured
✅ Monthly cost review meetings
✅ FinOps culture adoption

Sonuç

FinOps continuous process. Monthly review, optimize, repeat. Average %30-40 savings achievable. Devups FinOps assessment.