The 3-2-1 rule is the rare piece of IT doctrine your customers may already know: keep 3 copies of your data, on 2 different types of media, with 1 copy offsite. It survives because it's simple and because it maps to the three ways data actually dies: deletion and corruption, device failure, and site-level disaster. Here's how an MSP turns the slogan into an architecture — and into a sellable service.
Copy 1: production data
The live data on workstations, servers, and SaaS tenants counts as the first copy. The MSP job here is inventory: you can't protect the NAS nobody mentioned or the Microsoft 365 tenant everyone assumed "Microsoft backs up." (Microsoft replicates for availability; it does not protect your customer from deletion, ransomware, or a malicious insider. That's what M365 backup is for.)
Copy 2: local backup — the fast restore
An onsite copy — NAS, local disk, or a backup server — exists for one reason: speed. When a user deletes a folder or a disk dies, restoring over the LAN takes minutes instead of hours. Local-only backup, though, dies with the building: fire, flood, theft, and — most commonly — ransomware that encrypts the backup share right along with production. Local backup is your recovery-time story, never your survival story.
Copy 3: offsite — the survival copy
The offsite copy is what makes the strategy disaster-proof, and it's where a hosted platform carries the weight: encrypted transfer offsite, storage on infrastructure that is itself redundant (our triple-parity RAID-Z3 ZFS arrays survive three simultaneous drive failures — see our infrastructure), and versioning so ransomware that struck Tuesday can be rolled back to Monday. Two properties matter enormously here:
- Isolation — offsite copies live outside the customer's credentials and network, so a compromised domain admin can't reach them
- Immutability / versioning — point-in-time history, not a mirror that faithfully replicates the encryption event
The "two media" clause in 2026
The rule predates cloud storage; "two media types" once meant disk plus tape. The modern reading is two failure-independent systems: a local NAS and a geographically separate cloud platform qualify, because no single event — mechanical, electrical, or criminal — takes out both. Disk plus a second disk in the same chassis does not.
Deploying it as an MSP package
3-2-1 maps directly onto a sellable service tier:
- Agent on every workstation and server (files, images, databases, hypervisors)
- Local backup target for fast operational restores
- Encrypted offsite copy to your branded cloud, with retention configured per client
- Daily job monitoring from the admin panel, monthly branded report to the client
- A quarterly restore drill — the proof that turns the architecture into renewals
Because the platform is white label, the entire stack — software, portal, reports — carries your brand, and the 3-2-1 diagram in your proposal is a diagram of your service.
Selling the gap
The rule's best commercial feature is that most prospects fail it visibly. A five-minute audit usually finds local-only backups, an unprotected SaaS tenant, or a "backup" that's really a sync. Show the customer which of the three numbers they're missing, quote the tier that fixes it, and the conversation prices itself — packaging guidance in How MSPs Should Price Backup Services.