Backup & Catastrophe Restoration for Devoted Servers

Restoration Time Goal (RTO) is how lengthy your software could be offline earlier than the enterprise impression turns into unacceptable. A SaaS software with enterprise clients might need an RTO of half-hour. A advertising web site would possibly tolerate 4 hours.

Restoration Level Goal (RPO) is how a lot knowledge loss is appropriate. An ecommerce retailer taking 500 orders per hour can’t afford to lose 24 hours of knowledge — every misplaced order is misplaced income and a customer support downside. That very same retailer would possibly run backups each quarter-hour. A improvement server would possibly run weekly backups with an RPO of seven days.

Doc each numbers earlier than deciding on backup frequency, retention, and tooling. Your backup structure needs to be designed to fulfill outlined necessities, not inherited from no matter was configured at provisioning.

The three-2-1 Rule Utilized to Devoted Servers

The three-2-1 rule is the foundational framework for manufacturing backup technique:

• 3 copies of the info
• 2 completely different storage media varieties
• 1 copy saved offsite (geographically separate from the first server)

Utilized to a devoted server, this usually appears like:

1. Main knowledge: The reside manufacturing server itself
2. Native backup: An on-server backup to a separate disk or partition (completely different from the OS/software disk)
3. Offsite backup: A duplicate shipped to object storage (S3-compatible), a secondary devoted server, or InMotion’s backup storage

A single backup saved on the identical server as the info will not be a backup technique. Ransomware encrypts hooked up storage. Disk failures take down all the pieces on the identical controller. A neighborhood backup with out an offsite copy fails each assessments.

InMotion’s Premier Care bundle consists of 500GB of backup storage for devoted server clients, offering the offsite copy part with out requiring separate cloud storage accounts.

Restic

Restic is a contemporary, quick, encrypted backup device that helps S3, Backblaze B2, SFTP, and native storage as backup locations. Its deduplication and encryption are built-in, and the repository format is forward-compatible with future restic variations.

Primary backup configuration for an internet server:

# Initialize a repository on S3-compatible storage

restic -r s3:https://s3.amazonaws.com/your-bucket/server-backups init

# Set credentials through setting variables

export AWS_ACCESS_KEY_ID="your_key"

export AWS_SECRET_ACCESS_KEY="your_secret"

# Backup net root and database dumps

restic -r s3:https://s3.amazonaws.com/your-bucket/server-backups 

  backup /var/www /and so forth /dwelling 

  --exclude /var/www/html/cache 

  --exclude /var/www/html/tmp

# Apply retention coverage (preserve 7 day by day, 4 weekly, 6 month-to-month)

restic -r s3:https://s3.amazonaws.com/your-bucket/server-backups 

  neglect --keep-daily 7 --keep-weekly 4 --keep-monthly 6 --prune

Run through cron hourly or nightly relying in your RPO.

Duplicates

Duplicati supplies a web-based administration interface, which makes it accessible for server operators preferring to not handle backup jobs completely through command line. It helps encryption, compression, and a number of backend storage targets. For cPanel-managed servers, Duplicati can run as a system service backing up particular directories.

Rsync for Incremental Copies

rsync with –link-dest creates environment friendly incremental backups the place unchanged recordsdata are hard-linked to the earlier backup snapshot somewhat than duplicated. This supplies a number of point-in-time snapshots at storage value proportional to the info that modified:

#!/bin/bash

BACKUP_DIR="/backup/net"

DATE=$(date +%Y-%m-%d_percentH-%M)

LATEST="$BACKUP_DIR/newest"

rsync -avz --link-dest="$LATEST" /var/www/ "$BACKUP_DIR/$DATE/"

# Replace the symlink to newest backup

rm -f "$LATEST"

ln -s "$BACKUP_DIR/$DATE" "$LATEST"

Database-Particular Backup

File-system backups of working MySQL databases seize inconsistent states — the database recordsdata on disk could also be mid-transaction when the backup runs. Database backups should use database-native instruments:

MySQL/MariaDB:

# Full logical backup

mysqldump --all-databases --single-transaction --routines --triggers 

  -u root -p > /backup/mysql_$(date +%Ypercentmpercentd).sql

# Compress to scale back storage

gzip /backup/mysql_$(date +%Ypercentmpercentd).sql

For bigger databases the place mysqldump takes too lengthy, Percona XtraBackup supplies bodily sizzling backups that seize constant state with out locking tables and restore considerably sooner than logical dump recordsdata. Percona XtraBackup documentation covers set up and utilization.

PostgreSQL:

pg_dumpall -U postgres | gzip > /backup/postgres_$(date +%Ypercentmpercentd).sql.gz

cPanel Backup Supervisor

In case your devoted server runs cPanel/WHM, the built-in Backup Supervisor handles account-level backups together with recordsdata, databases, electronic mail, and configuration in a single operation. Configure it in WHM underneath Backup > Backup Configuration.

Key settings to evaluation:

• Backup Kind: Full and incremental (not simply full — incremental between full backups reduces RPO with out proportional storage value)
• Distant Vacation spot: Configure SFTP or S3-compatible distant vacation spot for the offsite copy
• Retention: Match to your documented RPO and storage price range

InMotion’s Backup Supervisor add-on supplies an extra 500GB of backup storage accessible instantly inside cPanel, with no extra server configuration required.

Testing Restores

A backup that has by no means been examined is a speculation, not a safety. Testing ought to cowl three situations:

File-level restore: Restore a particular listing from backup and confirm the contents match what was backed up. Do that month-to-month.

Database restore: Restore the latest database dump to a staging server and confirm software operate. This catches corruption within the dump file earlier than you want it in manufacturing. Do that month-to-month.

Full server restore drill: Provision a brand new server, restore all the pieces from backup, and confirm the appliance runs appropriately. That is the take a look at that reveals whether or not your documentation, backup configuration, and restore procedures really work end-to-end. Do that quarterly.

Doc the restore process explicitly, together with instructions, timing, and verification steps. The individual performing a restore throughout an incident is probably not the one who wrote the backup configuration.

Automating Backup Verification

Restic supplies a built-in integrity test that verifies backup repository consistency and not using a full restore:

restic -r s3:https://s3.amazonaws.com/your-bucket/server-backups test

Run this weekly through cron and alert on non-zero exit codes. A corrupted backup that is by no means checked is indistinguishable from a very good backup till you want it.

For file-level verification, add a checksum comparability step to your backup course of:

# Retailer a checksum of the supply listing

discover /var/www -type f -exec md5sum {} ; | type > /backup/checksums_$(date +%Ypercentmpercentd).txt

# Examine after restore

md5sum -c /backup/checksums_20260227.txt

Catastrophe Restoration Planning Past Backups

Backups restore knowledge. A catastrophe restoration plan addresses all the pieces else:

• DNS failover: In case your server is down, are you able to redirect site visitors to a brief alternative? Cloudflare’s proxied DNS makes IP-level failover a matter of adjusting an A document.
• Documented server configuration: Your backup comprises your knowledge, not your Nginx configuration, firewall guidelines, PHP-FPM pool settings, and software dependencies. Doc or automate server configuration so a brand new server could be provisioned to match the earlier one.
• Communication plan: Who will get notified when the server goes down? In what order? Who has authority to take the server offline for upkeep throughout a restoration?

InMotion Internet hosting’s Premier Care consists of entry to InMotion Options — the consulting staff that assists with complicated server restoration situations. The 1 hour/month of Options time is most valuably used to check catastrophe restoration procedures and doc restoration runbooks earlier than they’re wanted.