Off-site backup for SMEs - options, costs, planning, implementation
Why a local backup is not enough, what off-site options there are and what they cost - a fair comparison with planning help, worked examples and a checklist.
An off-site backup is an additional copy of your backup that is kept physically, network-wise and administratively separate from your business - so that fire, theft or ransomware cannot hit the original data and the backup at the same time. For most SMEs, an automated, encrypted and immutable copy to an external target - such as a cloud repository at a service provider or object storage - is more practical than rotating hard drives. Tape or a second site of your own pays off mainly with large data volumes or existing infrastructure.
This guide compares the common options, shows how to plan bandwidth, retention and recovery, and works through the costs in a transparent way. In the interest of transparency: we offer one of these options ourselves with Veeam Cloud Connect. That is why we also tell you when it is not the right fit.
What does ‘off-site’ mean - and why is a local backup not enough?
‘Off-site’ means ‘away from the site’: a copy that survives an incident at your main location. A backup server or NAS in the same server room protects you well against failed hard drives and accidentally deleted files, but it shares many risks with the original data:
- Fire and firefighting water: a fire in the server room hits servers and backup at the same time.
- Flooding and heavy rain: servers in the basement are particularly at risk.
- Theft: anyone who takes the servers usually takes the backup device next to them as well.
- Power surges: a lightning strike can damage all connected devices.
- Ransomware: attackers with administrator rights deliberately search the same network for backups and encrypt or delete them.
The last risk is the most pressing. In its 2025 status report, the BSI (Germany’s Federal Office for Information Security) counts around 950 publicly known ransomware attacks, mostly on SMEs. According to a Veeam study, the backup repositories were also targeted by the attackers at 89% of the affected organisations. The BSI therefore advises in its ransomware catalogue of measures to keep ‘at least one copy offline’.
The 3-2-1 rule sums up the principle: three copies, two different media, one copy at another location. Veeam extends it to 3-2-1-1-0 - one additional copy offline or immutable, zero errors during verification. Neither is a standard, but both are a good guide.
What do the BSI, the GDPR and cyber insurers require?
- BSI IT-Grundschutz CON.3: backup media must be stored physically separate from the systems backed up, ideally in a different fire compartment (A12). It must not be possible to overwrite backups intentionally or unintentionally (A14), and restoring them must be tested regularly (A15).
- Online backup (CON.3.A9): for online storage, aspects such as storage location, availability, authentication and encryption at rest and in transit should be governed by contract - and the network connection must match the permissible recovery times.
- BSI ransomware catalogue of measures: at least one offline copy, copies at different locations and a regular check of the separation: a domain administrator should try to access the separated copies as a test - and should not succeed.
- GDPR Art. 32 (EU General Data Protection Regulation): it must be possible to restore personal data promptly after a physical or technical incident, and the measures must be reviewed regularly.
- NIS-2: for companies covered by the new BSI Act, backup management and disaster recovery are part of risk management (Section 30 BSIG, the German Act on the Federal Office for Information Security). Find out more in our article NIS-2 and data backup.
Cyber insurers also take a close look: according to the BSI, insurance terms specify concrete measures such as a backup concept. The non-binding model conditions of the GDV, the German Insurance Association (AVB Cyber, as of February 2024), for example, call for a backup at least once a week in which no system and no authorisation level can alter or destroy both the original and the duplicate; the backup server should not be integrated into a management system such as Active Directory. What counts is your own policy.
How much distance and separation makes sense?
Distance alone is not enough. A copy can be far away and still be vulnerable if it is managed with the same credentials, as Veeam also points out. So think in four dimensions:
| Dimension | Key question | Good practice |
|---|---|---|
| Physical | Does the copy survive an incident at the site? | At least a different fire compartment, better a different building or location |
| Network | Can malware reach the copy over the network? | No permanently mapped drive, only defined connections |
| Administrative | Can a compromised admin account delete the copy? | Separate credentials outside Active Directory, ideally a different operator |
| Time | Can the copy be altered before a set period expires? | Immutability or offline media with sufficient retention |
There is no fixed distance in kilometres. The BSI requires physical separation and recommends a different fire compartment; the Bavarian State Office for Information Security (LSI) advises a different building, ideally a different location. Only a target outside the region protects against regional events such as flooding.
With ransomware, the time dimension is decisive: a copy that technically cannot be deleted before a set period expires also survives stolen credentials. Find out more in our article Immutable backups against ransomware.
What options are there for off-site backup?
This overview is deliberately simplified - how well an option works depends heavily on how it is implemented.
Rotating USB hard drives
- Automation
- low, manual rotation
- Network separation
- good, as long as disconnected
- Immutability
- offline, as long as disconnected
- Recovery
- fast, once the drive is on site
- Effort and sources of error
- high, forgotten rotations
- Cost structure
- low purchase cost
NAS at a second site
- Automation
- high
- Network separation
- often weak (same network, same domain)
- Immutability
- depends on the device, e.g. snapshots
- Recovery
- over the line or fetch the device
- Effort and sources of error
- medium, updates and permissions
- Cost structure
- hardware and line
Tape (LTO)
- Automation
- medium, rotation and transport
- Network separation
- very good, offline
- Immutability
- offline, WORM media possible
- Recovery
- slow for individual files
- Effort and sources of error
- high, handling and storage
- Cost structure
- drive, media, storage
Your own server in a data centre
- Automation
- high
- Network separation
- good with consistent hardening
- Immutability
- yes, e.g. hardened Linux repository
- Recovery
- over the line or fetch the server
- Effort and sources of error
- high, full operation
- Cost structure
- hardware, rack space, operation
Object storage in the public cloud
- Automation
- high
- Network separation
- good
- Immutability
- yes, with Object Lock
- Recovery
- over the internet
- Effort and sources of error
- medium, configuration and keys
- Cost structure
- usage-based, possibly retrieval and traffic
Veeam Cloud Connect at a service provider
- Automation
- high
- Network separation
- good, one outbound port
- Immutability
- can be added, depending on the offer
- Recovery
- over the internet
- Effort and sources of error
- low, monitor jobs
- Cost structure
- monthly price per device and storage
Rotating USB hard drives
Several hard drives are connected in rotation, with one always kept off-site. This is inexpensive and, as long as the drive is disconnected, a genuine offline backup. The weakness: the rotation depends on one person. During holidays, the drive often stays connected for weeks - and is reachable by ransomware. If someone takes it home, personal data leaves the company - encryption is then indispensable. According to the LSI, the storage location for backup media must also meet the requirements for a server room; a private household generally cannot do that. Useful as a supplement in very small environments, error-prone as the sole off-site strategy.
NAS at a second site or branch office
A NAS in the branch office or a second building receives copies automatically and is fast if the line is good. The catch is separation: it is often mapped as a network drive, a member of the same domain or reachable with the same admin accounts - physically separate, but not administratively. Remedies include separate credentials without domain membership, immutable snapshots (depending on the device) or a hardened repository instead of a file share. Hardware, updates and monitoring remain your responsibility.
Tape (LTO)
Tape is robust, offline and suitable for large data volumes. LTO-10 stores 30 TB natively; the LTO consortium has announced 40 TB cartridges for 2026. Important: according to the LTO consortium, LTO-10 drives only read and write LTO-10 media - for older tapes you will still need an older drive. Costs arise for the drive or tape library, media and secure storage, for example in a bank safe deposit box. On top of that comes manual work: rotating, labelling, transporting. With just a few terabytes, that is a lot of effort; with large data sets or existing tape infrastructure, tape remains a good offline option.
Your own server in a data centre (colocation)
You rent rack space in a data centre and run your own backup server there, such as a hardened Linux repository with immutability. This gives you full control, but also full responsibility for operation: purchase, rack space, power, connectivity, patches, hardware replacement, monitoring. The administrative separation is only as good as your access management. Sensible if you have your own IT team, larger data volumes or special control requirements.
Object storage in the public cloud (S3-compatible)
Veeam Backup & Replication and other backup programs write directly to S3-compatible object storage. With S3 Object Lock, backups become immutable for a defined period; in compliance mode, according to the AWS documentation, not even the root account can delete them early. The price is running it yourself: the bucket, Object Lock, access keys and permissions must be set up and maintained correctly. Depending on the provider, requests and outbound data traffic cost extra, with Amazon S3 for example above a monthly free allowance - relevant for large restores. Also check the storage location and jurisdiction.
Veeam Cloud Connect at a service provider
If you use Veeam, you can add the cloud repository of a Veeam Cloud & Service Provider as a target. The copy runs automatically over a single outbound port, without a VPN. The provider operates the infrastructure, while you keep control of jobs, retention and encryption. Depending on the offer, immutability and the ‘Insider Protection’ recycle bin protect against deletion. The limits: you need Veeam software, large restores are limited by your internet connection, and you tie yourself to a service provider - look for short terms and a clear exit. Our page on Veeam Cloud Connect shows how this works with us.
How does Veeam Cloud Connect work technically?
Veeam Cloud Connect is part of Veeam Backup & Replication. From your point of view, the setup works like this:
- The service provider creates a tenant account for you with a storage quota in the cloud repository.
- You add the provider as a service provider in the Veeam console: DNS name of the cloud gateway, port (default: 6180) and credentials. Your backup server checks the provider’s TLS certificate in the process.
- The cloud repository appears as a backup target in your console.
- A backup copy job automatically copies your local backups there. Backup jobs can also write directly to the cloud repository, but copying local backups is usually the better choice - you keep fast local recovery.
Your firewall only needs to allow the outbound connection to the cloud gateway on the agreed port - without a VPN tunnel or inbound rule. The copy is not a drive in your network that malware could simply overwrite.
What protection mechanisms are there?
- Encryption in transit: Veeam encrypts data traffic to and from the cloud repository by default.
- Your own key: in addition, you encrypt the backups in the job yourself (256-bit AES). Only you know the password - without the password, no restore is possible.
- Tenant separation: each tenant has its own area in the cloud repository and no access to other tenants’ data.
- Insider Protection: if someone deletes backups - even an attacker with your credentials - Veeam moves them to a recycle bin at the provider for a defined period. You cannot see it and cannot empty it; in an emergency, the provider makes the data available again.
- Immutability: if the cloud repository resides on a hardened repository, backups can neither be altered nor deleted for a defined period.
Which systems can back up to Cloud Connect?
- Veeam Backup & Replication: backup copy jobs copy the local backups of your backup server. Veeam supports direct backup jobs to the cloud for VMs from VMware vSphere and Microsoft Hyper-V.
- Veeam Agents for Windows, Linux and macOS: they can also back up on their own, without a backup server, directly to the cloud repository - but only in the Workstation or Server edition, not in the Free edition.
- Limitations: IBM AIX and Oracle Solaris are not supported, and restrictions apply to individual platforms - for Proxmox VE, for example, when restoring from backup copies. Check your platform in advance; find out more in our article Backup for Proxmox, Hyper-V and XCP-ng.
When is Cloud Connect a good fit - and when not?
Cloud Connect is a good fit if you use or are introducing Veeam, do not want to run a second site of your own and prefer predictable monthly costs. It is less suitable if you use other backup software, have very large data volumes over a weak connection, or need to be fully operational again within hours after a total failure - in that case you also need replacement hardware or replication. Statutory long-term retention is the job of an archiving system, not a backup.
Planning: from RPO and RTO to bandwidth
Define RPO and RTO
The RPO (Recovery Point Objective) states the maximum tolerable data loss, the RTO (Recovery Time Objective) how long recovery may take. Define both for each system together with the business departments; in its backup concept, the BSI provides for an RPO for each IT system. For the off-site copy, the RPO corresponds to the schedule of the copy job, usually daily. In the event of a total failure, the RTO depends above all on how quickly you can get the data back.
Calculate bandwidth and transfer time
Transfer time in hours = data volume in GB × 8,000 ÷ usable bandwidth in Mbit/s ÷ 3,600 (1 GB = 8,000 megabits, decimal units)
Do not calculate with the full nominal bandwidth; overhead and day-to-day operations need their share. Example with assumed values: upload 50 Mbit/s, 80% of which is usable (40 Mbit/s), first full backup after compression 700 GB, daily changes after compression 10 GB.
| Operation | Data volume (assumption) | Calculation | Duration (rounded) |
|---|---|---|---|
| First full backup | 700 GB | 700 × 8,000 ÷ 40 ÷ 3,600 | approx. 39 hours |
| Daily copy | 10 GB | 10 × 8,000 ÷ 40 ÷ 3,600 | approx. 33 minutes |
The daily copy fits comfortably into the night; the initial backup needs a weekend. To keep the line free during the day, you can limit bandwidth in Veeam based on time of day.
The first full backup: seeding
If the first transfer takes too long, seeding helps. According to Veeam, you create the first full backup of the backup copy job locally on a storage device and remove it from the job configuration. The provider copies the files into your area of the cloud repository, you rescan the repository and assign the backup to your job (‘mapping’) - after that, only changes are transferred. With us, seeding is possible by arrangement. It is rarely necessary: a weekend or a few nights are usually enough for the first transfer.
Incremental transfer and retention
After the initial backup, Veeam only transfers changed data. You define in the job how long restore points remain in the cloud repository, for example 30 daily points. For longer periods, there is GFS retention (Grandfather-Father-Son) with weekly, monthly or yearly full backups. Each level takes up storage, and with immutability enabled, data can only be deleted once the set period has expired. Plan in a buffer.
Plan realistically for recovery from the cloud
You can get individual files or emails back from the cloud repository relatively quickly. For entire servers, the line sets the pace: with an assumed 100 Mbit/s download, 80% of which is usable, the pure transfer of 700 GB takes around 19.5 hours (700 × 8,000 ÷ 80 ÷ 3,600). According to Veeam, Instant VM Recovery, i.e. starting a VM directly from the backup, is generally not supported from the cloud repository on the customer side, apart from a few platform-dependent exceptions.
The local backup therefore remains your tool for fast restores, and the off-site copy your safeguard against a total failure. Record in your emergency plan where you will restore to and over which line. Our article Testing restores shows how to practise this.
What does an off-site backup cost?
Market prices for hardware, rack space and cloud storage vary widely. Instead of flat figures, we show the cost components you should compare:
| Option | One-off | Ongoing | Often underestimated |
|---|---|---|---|
| Rotating USB hard drives | Hard drives | Replacement after a few years | Working time, forgotten rotations |
| NAS at a second site | NAS, hard drives | Power, line, maintenance | Updates, hardware replacement, room security |
| Tape (LTO) | Drive or library, media | Media, storage, transport | Handling, generation changes |
| Your own server in a data centre | Server, storage | Rack space, power, connectivity, maintenance | Operation, patches, on-site visits |
| Object storage in the public cloud | your own setup time | Storage per GB, possibly requests and data retrieval | Cost of large restores |
| Veeam Cloud Connect | your own setup time | Monthly price per device and storage | Bandwidth, buffer for data growth |
With every option, there is also internal working time for monitoring and restore tests.
Worked example: three servers at myveeam.cloud
A typical SME backs up three servers (VMs or physical servers) with Veeam. Including retention, around 1 TB accumulates in the cloud repository, in line with the example above: a 700 GB full backup plus around 30 daily changes of 10 GB each. With our prices, this results in:
| Item | Calculation | Monthly (net, plus VAT) |
|---|---|---|
| Basic package incl. 1 device and 200 GB | 1 × €39.90 | €39.90 |
| Two additional devices | 2 × €6.90 | €13.80 |
| 1 TB storage package | 1 × €36.00 | €36.00 |
| Total | €89.70 |
The storage packages are added to the 200 GB of the basic package: with 500 GB, the total would be 700 GB and therefore too tight; with the 1 TB package, 1.2 TB is available - around 200 GB of buffer for data growth. Over a year, that comes to €1,076.40 net plus VAT - with no minimum term, cancellable monthly, for businesses only. Not included are your local backup environment (backup server, storage, Veeam licences), your internet connection and optional extras such as immutability, Insider Protection or the 24/7 hotline; we will be happy to tell you the terms. Find out more on our page about Veeam Cloud Connect.
Data protection and law: what do you need to consider?
- Data processing on behalf: if you store backups with a service provider, this is processing on your behalf - you need a contract under Art. 28 GDPR.
- Storage location and jurisdiction: clarify the country and data centre, any subcontractors involved and the law to which the provider is subject.
- Your own key: for increased protection requirements, the BSI recommends encrypting all backups and storing the keys separately (CON.3.A13).
- A backup is not an archive: under Section 147 AO (German Fiscal Code), tax-relevant records must be retained for ten, eight or six years depending on their type - and, according to the GoBD (German principles for the proper keeping of electronic books and records), in complete, unalterable and machine-readable form. A backup serves recovery and is overwritten after its retention period; it does not replace an audit-proof archive.
- Deletion concept: retention periods for backups should match your deletion concept.
Note: This article provides a general overview and does not constitute legal advice. For binding statements, please contact your data protection officer, your tax adviser or a lawyer.
Checklist: what should you look for when choosing a provider?
- Location: country and data centre confirmed in writing, the provider’s jurisdiction clarified.
- Certification: for example ISO/IEC 27001 - and clarity on what the certificate covers.
- DPA: data processing agreement under Art. 28 GDPR, list of subcontractors.
- Encryption: in transit and with your own key that only you know.
- Protection against deletion: immutability and recycle bin available, retention period selectable.
- Access: separate credentials outside your Active Directory.
- Costs: transparent for devices and storage, fees for data retrieval clarified.
- Contract term and exit: short notice periods and a documented way back to your data.
- Large restores: what options are there if the line is too slow for a complete restore?
- Support: availability, language and response times, also at night and at weekends if required.
- Trial option: free trial with real data, including recovery.
Off-site backup in 5 steps
- Take stock: record systems, data volumes, daily change rate, existing backup software and actual upload bandwidth. Also think of data outside your servers, for example in Microsoft 365.
- Set objectives: RPO and RTO for each system, retention period, need for immutability, budget.
- Test: set up trial access, back up a representative system, measure real transfer times.
- Set up: backup copy jobs, encryption with your own password, retention, seeding if required, error notifications. Store the password and credentials separately from the backup server.
- Test recovery: restore one file and one complete server from the off-site copy, compare the time with your RTO and repeat the test regularly.
This is how the free trial at myveeam.cloud works: it lasts up to 30 days and always ends at the end of the month. During this time, you can run through steps 3 to 5 with your own data. Email us at hello@myveeam.cloud or call us on +49 8106 928 9493. If you continue afterwards, the monthly prices stated above apply, with no minimum term.
Frequently asked questions
Is synchronisation with cloud storage an off-site backup?
No. Synchronisation also transfers deletions and encrypted files. Versioning helps only to a limited extent and does not replace an independent copy with its own retention and its own credentials.
How far away should the off-site target be?
There is no fixed distance. At least a different fire compartment, better a different location - only a target outside the region protects against flooding.
Do I still need the local backup if I copy to the cloud?
Yes. The local backup enables fast restores in day-to-day operations, while the off-site copy protects against a total failure. Only both together meet the 3-2-1 rule.
Can the provider read my data?
If you encrypt the backups in the Veeam job with your own password, they are only stored at the provider in encrypted form. Only you know the password - keep it safe and separate.
Do I need Veeam Backup & Replication for Cloud Connect?
For backing up entire virtual machines, yes. Individual servers and workstations can also back up directly to a cloud repository with Veeam Agents for Windows, Linux or macOS, but only in the Workstation or Server edition.
Conclusion
An off-site backup belongs in every backup concept - not because of any single regulation, but because fire, theft and ransomware can also hit local backups. Separation is what matters: physical, network, credentials and time-based through immutability. USB drives, NAS, tape, colocation, object storage and Cloud Connect can all deliver this, but differ significantly in effort, susceptibility to error and cost structure. Plan bandwidth and recovery time realistically and test for an emergency before it happens.
We are a Veeam Cloud & Service Provider and a registered Veeam Value-Added Reseller. With Veeam Cloud Connect, you transfer your backup copies in encrypted form to our cloud repository in an ISO/IEC 27001 certified data centre in Germany, additionally with your own key if you wish. Our cloud repository is a Veeam Hardened Repository; immutability and Insider Protection can be added with a freely selectable retention period, and we provide the data processing agreement. If you would also like your on-site backup server to be monitored, our Veeam Managed Service takes care of this with individually agreed response times.
Sources
- BSI: IT-Grundschutz module CON.3 Backup Concept (Datensicherungskonzept) (2023 edition)
- BSI: Ransomware catalogue of measures (Maßnahmenkatalog Ransomware) (23 February 2022)
- BSI: The State of IT Security in Germany 2025 (11 November 2025)
- LSI Bayern: LSI-Info T#01 Data Backup (Backup and Recovery) (version 1.3, 10 February 2020)
- GDV: General Insurance Conditions for Cyber Risk Insurance (AVB Cyber) (as of February 2024)
- Regulation (EU) 2016/679 (GDPR), Art. 28 and 32 (27 April 2016)
- BSI Act (BSIG), Section 30 (in force since 6 December 2025)
- German Fiscal Code (Abgabenordnung), Section 147 Rules on the retention of records (accessed 5 October 2026)
- BMF (German Federal Ministry of Finance): GoBD, 2nd amendment (14 July 2025)
- Veeam Help Center: How Veeam Cloud Connect Backup Works (version 13, accessed 5 October 2026)
- Veeam Help Center: Ports (Veeam Cloud Connect) (version 13.1.1, accessed 5 October 2026)
- Veeam Help Center: Cloud Repository (29 July 2026)
- Veeam Help Center: Insider Protection (accessed 5 October 2026)
- Veeam Help Center: Data Encryption and Throttling (accessed 5 October 2026)
- Veeam Help Center: Encryption Standards (accessed 5 October 2026)
- Veeam Help Center: Tasks with Cloud Repository (accessed 5 October 2026)
- Veeam Help Center: Veeam Cloud Connect Backup - Limitations (24 September 2026)
- Veeam Help Center: Veeam Agent for Microsoft Windows - Backup to Veeam Cloud Connect Repository (accessed 5 October 2026)
- Veeam Help Center: Veeam Agent for Linux - Backup to Veeam Cloud Connect Repository (version 13.1.1, accessed 5 October 2026)
- Veeam Help Center: Veeam Agent for Mac - Backup to Veeam Cloud Connect Repository (accessed 5 October 2026)
- Veeam KB2264: Backup Copy Seeding and Mapping for Cloud Connect (24 October 2024)
- Veeam Help Center: Long-Term Retention Policy (GFS) (accessed 5 October 2026)
- Veeam Help Center: How Immutability Works (Object Storage) (accessed 5 October 2026)
- Veeam Blog: Backup copies that withstand attacks (30 September 2026)
- Veeam Blog: 3-2-1 Backup Rule (accessed 5 October 2026)
- AWS: Locking objects with Object Lock (accessed 5 October 2026)
- AWS: Amazon S3 Pricing (accessed 5 October 2026)
- LTO Program: LTO Generation Compatibility (accessed 5 October 2026)
- LTO Program: New 40 TB LTO-10 Cartridge Specifications (November 2025)