An offsite data center is a facility, separate from where your servers run, that keeps a copy of your data or a standby environment. It is the part of a backup strategy that protects against events no local copy survives: fire, flood, theft, power grid failures and ransomware that spreads through your network. This guide explains the options, how far is far enough, and what to check before choosing an offsite data center for your backups.
The problem: every local copy shares the same fate
Many companies have backups, but all of them are in the same building: a NAS in the server room, a second disk in the server, a backup server in the same rack. A fire, a flooded basement, a long power outage or an attacker with network access reaches all of them at once. Ransomware groups specifically look for backup servers and delete or encrypt them before encrypting production.
The consequence is that a single event becomes a total loss: no servers, no backups, no way to restore. Insurance policies and data protection regulators increasingly ask for evidence of offsite copies precisely because of this.
What "offsite" really means
A copy is offsite when a single physical event or a single compromised system cannot affect it and production together. That has two dimensions:
- Physical separation: a different building, ideally on a different power grid and outside the same flood or disaster zone.
- Logical separation: different credentials and, ideally, immutability, so an attacker who controls your servers cannot delete the offsite copy.
There is no universal minimum distance. Choose a location far enough that the regional risks you face (floods, storms, earthquakes, grid failures) are unlikely to hit both sites at the same time, while keeping latency and travel acceptable for your recovery plan.
Offsite data center options
| Option | How it works | Advantages | Drawbacks |
|---|---|---|---|
| Second own site | Backup servers in another office or branch | Full control | Rarely built to data center standards; you manage everything |
| Colocation | Your hardware in a rented rack in a professional data center | Power, cooling, security and connectivity included; full control of the equipment | Hardware purchase, remote hands costs, capacity planning |
| Cloud object storage | Backups sent to S3-compatible storage | No hardware, pay per use, immutability with object lock | Egress fees and restore time over the internet |
| Managed backup service | A provider runs the backup software and the offsite storage | Monitoring, retention and restores handled for you | Less control; check where data is stored and who holds the keys |
| Tape vaulting | Tapes transported to a secure vault | Offline by nature, very low cost per terabyte | Slow restores, logistics, handling errors |
How to evaluate an offsite data center
- Facility resilience: redundant power and cooling. The Uptime Institute Tier classification (I to IV) is a common reference for designs with increasing redundancy.
- Security and certifications: physical access control and audits such as ISO/IEC 27001 or SOC 2 reports.
- Data residency: in which country the data is stored. Laws such as Brazil's LGPD and the EU's GDPR govern international transfers of personal data, and some contracts require data to stay in a specific country.
- Encryption and key ownership: data should be encrypted before it leaves your servers, with keys the provider cannot read.
- Immutability: support for object lock or write-once retention, so backups cannot be deleted before their retention period.
- Restore path and costs: how fast you can get data back, whether you pay egress fees, and whether the provider can ship a disk for very large restores.
- Connectivity: bandwidth from your servers to the site and redundancy of the links.
Bandwidth: the hidden constraint
The first full copy to an offsite data center is the slowest. As a rule of thumb, at a sustained 100 Mbit/s, 1 TB takes about 22 hours to transfer; at 1 Gbit/s, about 2.2 hours. After the first copy, incremental and deduplicated backups send only changes, usually a small fraction of the total. Plan for:
- the initial seed (over the network outside business hours, or by shipping a disk);
- the daily change rate, which decides the bandwidth you need every night;
- the restore, which may need to bring back everything at once.
Our guide on how much cloud backup storage you need shows how to estimate change rate and capacity.
Backup copy or disaster recovery site?
An offsite data center can play two roles. As a backup location, it stores copies that you restore somewhere when needed. As a disaster recovery site, it also has compute ready to run your systems, so you can fail over in minutes or hours instead of days. The second is more expensive; choose based on your recovery time objective. The RPO vs RTO guide helps set that target, and the Proxmox disaster recovery guide shows an example for virtualized environments.
A practical setup for small and medium companies
- Local backup for fast restores (backup server or Proxmox Backup Server on-site).
- Encrypted, incremental copy to an offsite data center or S3-compatible storage every night.
- Object lock or a separate account with credentials that production servers do not have.
- Monthly restore test from the offsite copy, not only from the local one.
This is the core of the 3-2-1-1-0 rule, and it covers the vast majority of disaster scenarios at a predictable cost. See also cloud backup vs local backup.
Backnix is being built to deliver exactly that offsite layer: encrypted backups of servers, databases, Docker and Proxmox sent to offsite S3-compatible storage. See the cloud storage page for the current status and the plans and prices.