How to cut your AWS RDS costs by 65% in 20 minutes (without touching production)

This is the single most common finding in every cloud bill I audit.

Dev and staging RDS instances running 24/7 at production size.

A db.r5.xlarge costs $876/month running continuously. If your team uses it 8 hours a day on weekdays, you're paying for 720 hours and using 170. That's a 76% waste rate on a single instance.

The fix takes 20 minutes and touches nothing in production.

---

The problem in numbers

db.r5.xlarge on-demand:  $876/month  (720 hrs)
db.r5.xlarge scheduled:  $306/month  (252 hrs — weekdays 8am-7pm)
Monthly saving:          $570/month
Annual saving:           $6,840/year

Multiply that by 2-3 dev/staging environments and you're looking at $15,000–$20,000/year on instances that sleep most of the time.

---

Step 1: Find your idle RDS instances

# List all RDS instances with their sizes and status
aws rds describe-db-instances \
  --query 'DBInstances[*].{ID:DBInstanceIdentifier,Class:DBInstanceClass,Status:DBInstanceStatus,MultiAZ:MultiAZ}' \
  --output table

Look for anything that:

• Has "dev", "staging", "test", or "qa" in the name
• Is the same instance class as production
• Has Multi-AZ enabled (almost never needed for dev)

---

Step 2: Create the IAM role

cat > trust-policy.json << 'EOF'
{
  "Version": "2012-10-17",
  "Statement": [{
    "Effect": "Allow",
    "Principal": { "Service": "lambda.amazonaws.com" },
    "Action": "sts:AssumeRole"
  }]
}
EOF

aws iam create-role \
  --role-name rds-scheduler-role \
  --assume-role-policy-document file://trust-policy.json

aws iam attach-role-policy \
  --role-name rds-scheduler-role \
  --policy-arn arn:aws:iam::aws:policy/AmazonRDSFullAccess

---

Step 3: Create the Lambda function

# rds_scheduler.py
import boto3
import os

def handler(event, context):
    rds = boto3.client('rds')
    action = event.get('action')  # 'start' or 'stop'
    instances = os.environ.get('RDS_INSTANCES', '').split(',')

    for instance_id in instances:
        if not instance_id.strip():
            continue
        try:
            if action == 'stop':
                rds.stop_db_instance(DBInstanceIdentifier=instance_id.strip())
                print(f'Stopped: {instance_id}')
            elif action == 'start':
                rds.start_db_instance(DBInstanceIdentifier=instance_id.strip())
                print(f'Started: {instance_id}')
        except Exception as e:
            print(f'Error on {instance_id}: {e}')

    return {'status': 'done', 'action': action, 'instances': instances}

Deploy it:

zip scheduler.zip rds_scheduler.py

aws lambda create-function \
  --function-name rds-scheduler \
  --runtime python3.12 \
  --role arn:aws:iam::YOUR_ACCOUNT_ID:role/rds-scheduler-role \
  --handler rds_scheduler.handler \
  --zip-file fileb://scheduler.zip \
  --environment "Variables={RDS_INSTANCES=your-dev-db,your-staging-db}"

---

Step 4: Create the EventBridge schedules

# Stop at 7pm every weekday
# Adjust UTC offset for your timezone — CET/Warsaw is UTC+1, so 18:00 UTC = 7pm local
aws events put-rule \
  --name "StopDevRDS" \
  --schedule-expression "cron(0 18 ? * MON-FRI *)" \
  --state ENABLED

# Start at 8am every weekday
aws events put-rule \
  --name "StartDevRDS" \
  --schedule-expression "cron(0 7 ? * MON-FRI *)" \
  --state ENABLED

# Get your Lambda ARN
LAMBDA_ARN=$(aws lambda get-function \
  --function-name rds-scheduler \
  --query 'Configuration.FunctionArn' \
  --output text)

# Add targets
aws events put-targets \
  --rule StopDevRDS \
  --targets "Id=1,Arn=$LAMBDA_ARN,Input='{\"action\":\"stop\"}'"

aws events put-targets \
  --rule StartDevRDS \
  --targets "Id=1,Arn=$LAMBDA_ARN,Input='{\"action\":\"start\"}'"

# Allow EventBridge to invoke Lambda
aws lambda add-permission \
  --function-name rds-scheduler \
  --statement-id allow-eventbridge \
  --action lambda:InvokeFunction \
  --principal events.amazonaws.com \
  --source-arn $(aws events describe-rule --name StopDevRDS --query 'Arn' --output text)

---

Step 5: Downsize the instance too

While you're here — dev doesn't need a db.r5.xlarge. Drop to db.t3.medium for another 70% saving:

aws rds modify-db-instance \
  --db-instance-identifier your-dev-db \
  --db-instance-class db.t3.medium \
  --apply-immediately

Combined result:

db.r5.xlarge 24/7:       $876/month  (original)
db.r5.xlarge scheduled:  $306/month  (schedule only)
db.t3.medium scheduled:  $89/month   (schedule + downsize)

Annual saving:           $9,444/year — from one dev database

---

Step 6: Add a wake-up Slack command (optional)

Your team will occasionally need the DB outside hours. Give them a self-service way to start it:

# Create a Lambda function URL
aws lambda create-function-url-config \
  --function-name rds-scheduler \
  --auth-type NONE

Then in Slack: Apps → Slash Commands → /start-devdb → point to the URL with body {"action": "start"}.

Engineers can start the DB themselves without waiting for someone with AWS console access.

---

What this looks like in real teams

I've helped teams set this up in about 20 minutes. The conversation afterwards is always the same:

"I can't believe we were paying for that for 14 months."

Not because they didn't know. Because nobody sat down and looked.

This is one of 18 checks in KloudAudit — a free self-guided cloud cost audit that walks through all of these systematically. No AWS credentials required. Takes 15 minutes.

---

Quick reference

Action	Command
Find idle RDS	aws rds describe-db-instances --query ...
Stop instance now	aws rds stop-db-instance --db-instance-identifier NAME
Start instance now	aws rds start-db-instance --db-instance-identifier NAME
Check instance state	aws rds describe-db-instances --db-instance-identifier NAME --query 'DBInstances[0].DBInstanceStatus'

---

What's your team's current RDS setup — scheduled or running 24/7?

Drop your setup in the comments — genuinely curious how common this is.