Infrastructure 06.2 — Backup

Proxmox Backup Server and a recovery plan

A backup is not what gets created — it is what gets restored. I deploy PBS, write the recovery plan and test it monthly, so that at moment X everything really comes back.

3-2-1
Strategy
30+days
History depth
1/mo
Restore test

Backup / capabilities

What it covers

A 3-2-1 strategy, deduplication, incremental copies and an offsite copy on S3 — in one manageable system.

Proxmox Backup Server

Deduplication, incremental copies and verification — the same volume takes far less space.

VMs, LXC and files

A whole VM, a container or a single folder — from one panel, on one schedule.

Offsite copy

A second PBS at another site or an S3 bucket remotely — fire, theft or ransomware cannot kill everything at once.

Schedule and retention

Daily, weekly and monthly copies — with retention periods set in writing.

Restore testing

Every month I restore a real copy into a separate environment and record the time — that is the real guarantee.

Documentation

What is stored where, how it is restored and who has access — in writing, not in someone’s head.

process

How the process runs

  1. 01

    Inventory

    What data exists, what is critical and what loss is acceptable.

  2. 02

    Strategy

    RPO and RTO — how much data may be lost and how fast it must come back.

  3. 03

    Build

    PBS, storage, schedule, offsite replication and access control.

  4. 04

    Test

    A restore rehearsal and a report — before you actually need it.

cases

Examples

Recovery

After ransomware

An encrypted server was restored from the previous day’s copy — operations resumed the same day.

Archive

12 TB archive offsite

Thanks to deduplication, 12 TB is kept offsite at a reasonable cost.

faq

Common questions

How often are backups taken?

Daily by default, every 6 hours for critical systems. The schedule matches your rhythm.

Where is the offsite copy kept?

At a second site or in an S3 bucket, encrypted. The key stays with you.

How fast does a server come back?

A small VM takes 10–20 minutes, a large system 1–2 hours. The exact figure is recorded after testing.

Do you know how fast your server would come back?

If you do not know — that is exactly why it is worth fixing. Let us start with an audit.

Let’s talk

Contact

Get in touch about your technical task

Contact me by phone or WhatsApp. Critical incidents are covered 24/7; other requests are handled within the agreed support schedule.

24/7 on critical incidents — Response mode

If I miss your call, I will get back to you.