Automation and Tooling¶

📖 FinOps Capabilities - All FinOps capabilities

Overview¶

This document covers FinOps automation, tooling, and engineering practices for implementing cost management at scale. Automation reduces manual effort and ensures consistent cost governance across cloud environments.

Cost Automation Patterns¶

Automated Tagging¶

# AWS Lambda function to auto-tag untagged EC2 instances
import boto3

def lambda_handler(event, context):
    ec2 = boto3.client('ec2')

    # Find instances without required tags
    instances = ec2.describe_instances(
        Filters=[{'Name': 'tag-key', 'Values': ['team']}]
    )

    all_instances = ec2.describe_instances()

    for reservation in all_instances['Reservations']:
        for instance in reservation['Instances']:
            tags = {t['Key']: t['Value'] for t in instance.get('Tags', [])}
            if 'team' not in tags:
                # Tag with default and alert
                ec2.create_tags(
                    Resources=[instance['InstanceId']],
                    Tags=[
                        {'Key': 'team', 'Value': 'unassigned'},
                        {'Key': 'needs-tagging', 'Value': 'true'}
                    ]
                )

Scheduled Resource Management¶

# Start/stop development instances on schedule
# AWS CloudWatch Events + Lambda or AWS Instance Scheduler

# Pattern:
# 1. Tag dev resources: schedule=business-hours
# 2. Lambda runs on cron: stop at 7 PM, start at 7 AM
# 3. Skip weekends
# 4. Savings: ~65% on dev/test compute

# AWS:
# - AWS Instance Scheduler (official solution)
# - CloudWatch Events + Lambda (custom)

# Azure:
# - Azure Automation with Start/Stop VMs
# - Azure DevTest Labs auto-shutdown

# GCP:
# - Cloud Scheduler + Cloud Functions
# - Instance schedules (native)

Automated Right-Sizing¶

# Right-sizing automation workflow:
1. Collect metrics (CloudWatch, Azure Monitor, GCP Monitoring)
   - CPU utilization (14-day average and P95)
   - Memory utilization
   - Network I/O

2. Generate recommendations
   - AWS Compute Optimizer API
   - Azure Advisor API
   - GCP Recommender API

3. Create tickets (Jira, ServiceNow)
   - Include current size, recommended size, estimated savings
   - Assign to resource owner

4. Implement (with approval)
   - Automated for non-production
   - Approval workflow for production

5. Validate
   - Monitor performance post-change
   - Rollback if issues detected

Budget Alerts and Enforcement¶

# AWS Budget with actions
{
  "BudgetName": "team-platform-monthly",
  "BudgetLimit": {
    "Amount": "10000",
    "Unit": "USD"
  },
  "BudgetType": "COST",
  "TimePeriod": {
    "Start": "2024-01-01",
    "End": "2087-06-15"
  },
  "TimeUnit": "MONTHLY",
  "CostFilters": {
    "TagKeyValue": ["user:team$platform"]
  },
  "NotificationsWithSubscribers": [
    {
      "Notification": {
        "NotificationType": "ACTUAL",
        "ComparisonOperator": "GREATER_THAN",
        "Threshold": 80,
        "ThresholdType": "PERCENTAGE"
      },
      "Subscribers": [
        {"SubscriptionType": "EMAIL", "Address": "platform-leads@company.com"},
        {"SubscriptionType": "SNS", "Address": "arn:aws:sns:us-east-1:123456:budget-alerts"}
      ]
    }
  ]
}

Infrastructure as Code (IaC) for FinOps¶

Terraform Cost Estimation¶

# Use Infracost for Terraform cost estimation
# infracost breakdown --path=.

# Example: estimate cost before applying
resource "aws_instance" "web" {
  ami           = "ami-12345678"
  instance_type = "m5.xlarge"  # $0.192/hr = ~$140/month

  tags = {
    team        = "platform"
    environment = "production"
    cost-center = "CC-1001"
  }
}

# Infracost output:
# Name                 Monthly Cost
# aws_instance.web     $140.16
# Total                $140.16

📖 Infracost - Terraform cost estimation

Policy as Code¶

# Open Policy Agent (OPA) / Cloud Custodian examples

# Cloud Custodian: Find untagged resources
# policies:
#   - name: untagged-ec2
#     resource: aws.ec2
#     filters:
#       - "tag:team": absent
#     actions:
#       - type: notify
#         subject: "Untagged EC2 Instance Found"
#         to: ["finops@company.com"]

# Cloud Custodian: Stop idle instances
# policies:
#   - name: stop-idle-instances
#     resource: aws.ec2
#     filters:
#       - type: metrics
#         name: CPUUtilization
#         statistics: Average
#         period: 604800  # 7 days
#         value: 5
#         op: less-than
#     actions:
#       - type: stop

📖 Cloud Custodian - Policy engine documentation

CI/CD Cost Gates¶

# GitHub Actions example: cost estimation in PR
name: Cost Estimation
on: pull_request
jobs:
  infracost:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: infracost/actions/setup@v3
        with:
          api-key: ${{ secrets.INFRACOST_API_KEY }}
      - run: infracost breakdown --path=. --format=json --out-file=/tmp/infracost.json
      - uses: infracost/actions/comment@v3
        with:
          path: /tmp/infracost.json
          behavior: update

Cloud Provider Cost APIs¶

AWS Cost Explorer API¶

import boto3

client = boto3.client('ce')

# Get cost and usage
response = client.get_cost_and_usage(
    TimePeriod={
        'Start': '2024-01-01',
        'End': '2024-02-01'
    },
    Granularity='DAILY',
    Metrics=['BlendedCost', 'UnblendedCost', 'UsageQuantity'],
    GroupBy=[
        {'Type': 'DIMENSION', 'Key': 'SERVICE'},
        {'Type': 'TAG', 'Key': 'team'}
    ]
)

# Get right-sizing recommendations
compute_optimizer = boto3.client('compute-optimizer')
recommendations = compute_optimizer.get_ec2_instance_recommendations()

📖 AWS Cost Explorer API - Cost API

Azure Cost Management API¶

# Azure Cost Management REST API
import requests

url = "https://management.azure.com/subscriptions/{sub_id}/providers/Microsoft.CostManagement/query"
headers = {"Authorization": f"Bearer {token}"}
body = {
    "type": "ActualCost",
    "timeframe": "MonthToDate",
    "dataset": {
        "granularity": "Daily",
        "aggregation": {
            "totalCost": {"name": "Cost", "function": "Sum"}
        },
        "grouping": [
            {"type": "Dimension", "name": "ServiceName"}
        ]
    }
}

📖 Azure Cost Management API - Cost API

GCP Cloud Billing API¶

-- BigQuery billing export query
SELECT
  service.description AS service,
  SUM(cost) AS total_cost,
  SUM(usage.amount) AS total_usage
FROM `project.dataset.gcp_billing_export_v1_XXXXXX`
WHERE invoice.month = '202401'
GROUP BY service
ORDER BY total_cost DESC;

📖 GCP Billing API - Billing API

Monitoring and Alerting¶

Cost Anomaly Detection¶

Provider	Service	Configuration
AWS	Cost Anomaly Detection	Service-level or account-level monitors
Azure	Cost alerts with anomaly detection	Budget-based anomaly alerts
GCP	Budget alerts	Threshold-based with forecasting

Alert Strategy¶

Tier 1 (Immediate): > 50% daily increase - page on-call
Tier 2 (Urgent): > 25% daily increase - Slack notification to team
Tier 3 (Informational): > 10% weekly trend - weekly report
Tier 4 (Forecast): Projected to exceed budget - monthly review

Automation Maturity¶

Crawl¶

Manual cost reports (spreadsheets)
Basic budget alerts via email
Ad-hoc tagging

Walk¶

Automated cost dashboards
Scheduled start/stop for dev environments
Automated tagging enforcement
Right-sizing recommendations reviewed monthly

Run¶

CI/CD cost gates blocking expensive deployments
Automated right-sizing with approval workflows
Real-time anomaly detection and response
Policy-as-code enforcement for all resources
Commitment purchasing automation