Disaster Recovery Planning for Small Teams

The 3-2-1 Rule

• 3 copies of your data
• 2 different storage types
• 1 offsite location

Example:

1. Production server (primary)
2. Local backup on separate disk (second copy, different media)
3. Remote backup at different provider (offsite)

Recovery Time Objectives

Tip Your RTO and RPO determine your backup strategy and cost. Real-time replication for RPO=0 is expensive. Daily backups for RPO=24h are cheap. Choose based on business impact.

What to Back Up

Critical (back up hourly or more)

• Database (full + WAL/binlog for point-in-time)
• User-uploaded files
• Application secrets and certificates

Important (back up daily)

• Application code (also in git)
• Server configuration files
• Cron jobs and scripts

Nice to Have (back up weekly)

• Log files
• Monitoring data
• Development environments

Automated Backup Script

#!/bin/bash
# /opt/backup/disaster-recovery.sh

set -euo pipefail

REMOTE="backup-server:/backups/$(hostname)"
DATE=$(date +%Y%m%d_%H%M)
LOCAL="/backup/$DATE"

mkdir -p "$LOCAL"

# 1. Database
echo "[$(date)] Backing up database..."
mysqldump --all-databases --single-transaction | gzip > "$LOCAL/mysql.sql.gz"

# 2. Application files
echo "[$(date)] Backing up app files..."
tar czf "$LOCAL/www.tar.gz" -C /var/www html --exclude='node_modules' --exclude='.git'

# 3. Configuration
echo "[$(date)] Backing up config..."
tar czf "$LOCAL/etc.tar.gz" /etc/nginx /etc/letsencrypt /etc/systemd/system/*.service 2>/dev/null

# 4. Push to remote
echo "[$(date)] Syncing to remote..."
rsync -az --delete "$LOCAL/" "$REMOTE/$DATE/"

# 5. Cleanup local (keep 3 days)
find /backup -maxdepth 1 -type d -mtime +3 -exec rm -rf {} \;

echo "[$(date)] Backup complete: $LOCAL"

Recovery Runbook

Document these steps BEFORE you need them:

1. Provision new server — same plan, same region
2. Install base packages — restore /etc configs
3. Restore database — import latest dump
4. Restore application files — extract www archive
5. Restore SSL certificates — or re-issue via Let's Encrypt
6. Update DNS — point domain to new server IP
7. Verify — check all endpoints, test login, verify data

Testing Recovery

Schedule a quarterly drill:

# Spin up a test server
# Restore from backup
# Verify everything works
# Document issues found
# Destroy test server

Danger An untested backup is not a backup. If you've never practiced a full restore, you don't know if it works, how long it takes, or what steps you'll forget under pressure.