Start with RTO and RPO, not the tool
Before choosing a backup product, define recovery objectives per workload. RPO (how much data you can afford to lose) drives backup frequency and replication. RTO (how fast you must be back) drives your recovery method — file-level restore, instant VM recovery, or full site failover. A clinical scheduling system may need RPO near zero and RTO in minutes; a file server may tolerate a 4-hour RPO and an 8-hour RTO. These numbers drive every architectural decision.
| Workload | RPO target | RTO target | Recovery method |
|---|---|---|---|
| Clinical/EHR | Near zero | Minutes | Replicated VM, instant recovery |
| Exchange/DAG | 15 min | 1 hour | Database copy activation |
| File server | 4 hours | 8 hours | File-level restore |
| M365 mailboxes | Daily | 4 hours | Third-party M365 backup |
| Archive | Weekly | Days | Cloud restore |
The 3-2-1 rule and immutability
Three copies of data, on two different media, with one off-site. In 2026 that means production, a fast on-prem backup target, and an immutable cloud copy. Immutability is non-negotiable against ransomware — a backup a ransomware operator can encrypt is not a backup. Use Object Lock / WORM on S3, Azure immutable blobs, or a dedicated immutable repository (Veeam Hardened Repository, Datto Immutable Cloud).
- Three copies, two media, one off-site — still the baseline
- At least one copy immutable (Object Lock / WORM / hardened repo)
- Air-gap the offline copy where feasible
- Test restores quarterly — an untested backup is a gamble
Veeam Backup & Replication
Veeam covers Hyper-V, VMware and physical agents with instant VM recovery, verified backups and SureReplica testing. It is the workhorse for most MSP estates. Configure a Scale-Out Backup Repository with a hardened Linux repo for the immutable tier, and use SureBackup to auto-verify each backup by booting it in an isolated network.
Import-Module Veeam.Backup.PowerShell
$job = Get-VBRJob -Name "VMs-Prod-Tier1"
Start-VBRJob -Job $job -FullBackup
# Verify the latest restore point
Get-VBRRestorePoint -Job $job | Select -Last 1 |
Format-List Name, CreationTime, Type, AlgorithmMSP360 / Cloudberry, Nakivo, Datto and Acronis
MSP360 (Cloudberry) excels at cloud-targeted file and image backups with granular retention and per-GB cloud cost control. Nakivo is a strong virtualised-estate option with built-in backup verification and site recovery orchestration. Datto BCDR appliances add ransomware detection and instant local virtualisation of failed servers from a backup. Acronis Cyber Protect couples backup with anti-malware in one agent, useful where a single vendor is preferred.
| Product | Strength | Best fit |
|---|---|---|
| Veeam | Broad hypervisor support, instant recovery | Mixed Hyper-V/VMware estates |
| MSP360/Cloudberry | Cloud-targeted, granular retention | File/image to S3/Azure/GCP |
| Nakivo | VM backup + site recovery orchestration | Virtualised estates |
| Datto BCDR | Appliance + ransomware detection | SMB DR with instant local VM |
| Acronis Cyber Protect | Backup + anti-malware in one agent | Single-vendor consolidation |
Microsoft 365 backup — do not assume Microsoft protects you
A common and costly misconception is that Microsoft backs up M365 data. Microsoft provides service availability, not long-term data protection — a deleted mailbox is gone after the retention window, and ransomware or a malicious admin can purge data fast. Add a third-party M365 backup (Veeam for M365, Datto SaaS, or equivalent) for Exchange, OneDrive, SharePoint and Teams.
RTO/RPO planning and implementation
Translate objectives into configuration: backup frequency = RPO, replication + instant recovery = RTO. Document the DR runbook per workload with the recovery steps, responsible role and target time. Keep a printed copy of the runbook — the DR event may take the wiki down. Run a full restore drill quarterly and time it; the drill is the only proof the plan works.
- Map each workload to RPO/RTO and recovery method
- Document the DR runbook per workload, printed copy off-site
- Run a full restore drill quarterly and time the recovery
- Include a ransomware scenario in the drill — restore to a clean network
Lessons learned
- Define RPO/RTO per workload before choosing a product
- Make at least one backup copy immutable
- Add a third-party M365 backup — Microsoft does not protect your data
- An untested backup is a gamble — drill quarterly and time it
Written by
Jamshed Alam
Global MSP & IT Infrastructure Consultant with 20 years across Exchange, Microsoft 365, Azure, security and automation. Available for project and contract work worldwide.