Skip to content
Back to Blog
Blog·May 2026·5 min read

The 3-2-1 backup rule, applied to Airtable.

Three copies. Two storage classes. One offsite. The 60-year-old rule that still decides whether your Airtable data survives a bad day — and where AlreadyBack fits.

What 3-2-1 actually means

The 3-2-1 rule was coined by photographer Peter Krogh in 2005 and has quietly become the baseline that auditors, cyber-insurance underwriters, and every serious IT shop reach for when they ask “is your data really backed up?”

  • 3 copies of your data in total
  • 2 different storage classes (or providers)
  • 1 copy offsite, geographically separated

It assumes nothing about whichfailure modes you're defending against. It just guarantees that no single failure — disk, datacenter, account, region, vendor — takes everything down with it.

Why this rule still applies in 2026 — even for SaaS-first teams

The objection sounds reasonable: “Airtable is in the cloud, isn't that already enough copies?”

The honest answer is no, for four concrete reasons:

  1. Airtable's trash is 7 days on most plans. If a deletion or bulk-update mistake is caught on day 8, the data is gone.
  2. Automations and scripts can move fast. A misconfigured automation flow can purge thousands of records in seconds. The Airtable trash captures the deletes; it does not capture the prior values of fields that got overwritten.
  3. Account compromise propagates.If an attacker gets a workspace owner's session, they can disable the trash, delete bases, and roll your account back through Airtable's own UI.
  4. Vendor incidents do happen. Airtable is excellent operationally, but no SaaS has a 100% record. Without a copy outside their tenant, an outage means waiting.

The 3-2-1 rule wasn't designed for SaaS, but the failure modes it defends against are exactly the ones modern teams now face.

Mapping 3-2-1 to a real Airtable setup

Here's what compliant looks like, concretely:

CopyWhere it livesWhat it defends against
#1 — LiveAirtable workspaceDay-to-day work; most failures don't reach it
#2 — Hot snapshotAlreadyBack encrypted store (US or EU, your choice)Bulk deletes, bad scripts, lost trash window, account compromise
#3 — Cold offsiteCross-cloud archive (second provider, different jurisdiction)Vendor-level outage, compromised AlreadyBack tenant, regional disaster

Two storage classes (Airtable proprietary + object storage), one offsite copy (cross-cloud), three independent copies. That's the rule, satisfied.

Where AlreadyBack handles the heavy lifting

You keep the live copy in Airtable. We handle the other two:

  • Copy #2 — Daily snapshots to our encrypted hot store, with US or EU data residency (your choice). AES-256-GCM at rest, TLS 1.3 in transit, optional BYOK on Resilience.
  • Copy #3— Cross-cloud cold archive. Today that's an automatic mirror to a second provider in a different jurisdiction. Soon: BYOS (Bring Your Own Storage), so the offsite copy can land in a bucket you control.
  • On the roadmap: Vault— WORM-locked snapshots, immutable at the storage layer, so not even a compromised account can purge its own backups. That covers the “ransomware on the backup itself” scenario every cyber-insurance underwriter now asks about.

TL;DR

If Airtable is operational data — the kind that moves billings, manages projects, or tracks regulatory state — you owe yourself the same backup discipline you'd apply to a database. The 3-2-1 rule is the cheapest insurance ever invented. Implement it once, sleep better forever.