Backup strategies: 3-2-1 rule and automated verification

Backups exist but have never been tested. When data loss occurs, the restore process fails or restores corrupt data.

Follow the 3-2-1 rule and automate restoration testing:

#!/bin/bash
# Automated Postgres backup with verification
DATE=$(date +%Y%m%d_%H%M%S)
BACKUP_FILE="/backups/db_$DATE.sql.gz"

# Create backup
pg_dump $DATABASE_URL | gzip > $BACKUP_FILE

# Verify backup is readable and non-empty
if [ ! -s $BACKUP_FILE ]; then
    echo "BACKUP FAILED: empty file" >&2
    exit 1
fi

GZ_TEST=$(gzip -t $BACKUP_FILE 2>&1)
if [ $? -ne 0 ]; then
    echo "BACKUP FAILED: corrupt gzip: $GZ_TEST" >&2
    exit 1
fi

# Upload to offsite storage (S3)
aws s3 cp $BACKUP_FILE s3://$BACKUP_BUCKET/db/

# Delete local backups older than 7 days
find /backups -name 'db_*.sql.gz' -mtime +7 -delete

echo "Backup complete: $BACKUP_FILE"

3-2-1 rule:

• 3 copies of data
• 2 different storage media/services
• 1 offsite copy

An untested backup is not a backup. Backups that fail silently (empty files, corrupt archives) are worse than no backup because they create false confidence.