Governing policy · POL-07

Business Continuity & Disaster Recovery Policy

Version 1.0 · effective 13 July 2026 · owner: Jonny Allum

OwnerJonny Allum
Version1.0 · Effective 13/07/2026 · Review annually (July) and after any DR test
Applies toThe BizOS production service and the business's ability to operate

1. What must survive

AssetWhyWhere it lives
Tenant dataCustomers' entire operations — the crown jewelsSupabase (managed Postgres, provider backups) — see SOP-06
Source code + migrationsThe platform can be rebuilt from repo + migrations aloneGitHub (jonnyallum/biz-os), local clones
SecretsRebuild is impossible without themPassword manager + VM env; register in SOP-09
Billing stateWho pays whatStripe (system of record for subscriptions)
DNS/domainsThe front doorRegistrar; records documented in SOP-06 §5

2. Objectives

  • RPO (data loss tolerance): 24 hours — bounded by Supabase's daily backup tier; anything tighter is a paid upgrade decision recorded here.
  • RTO (time to restore service): 1 working day for full service on a rebuilt VM; status communication to affected tenants within 4 hours (SOP-05 comms).

3. Continuity scenarios and stance

ScenarioStance
VM loss (GCP zone/instance failure)Rebuild from scratch: new VM, clone repo, restore env from password manager, pnpm install && build, PM2 + Caddy, repoint DNS. Rehearsed via SOP-06 §4.
Supabase incidentProvider status page monitored; restore from provider backup into a new project if catastrophic; app repointed via env.
Key person unavailable (sole operator)See §4 — the honest single point of failure.
Supplier exit (any of POL-09 register)Every store is pluggable (local/Supabase behind one contract); code and data are exportable; no supplier holds the only copy of anything.
Loss of workstationNo unique state on the endpoint: repo on GitHub, secrets in password manager, docs in repo. Re-provision and revoke old sessions same day.

4. The sole-operator contingency

The largest business continuity risk is the operator. Mitigations:

  1. Everything is written down — this governance pack, BUILD_PLAN.md's RESUME HERE handover block, and SOPs are sufficient for a competent engineer to operate the platform cold.
  2. A sealed continuity note (password-manager emergency access + instructions naming a trusted technical contact and next of kin) exists outside the business's own systems, so paying tenants are not stranded if the operator is incapacitated.
  3. Customer money is protected: subscriptions can be paused/refunded from Stripe by the emergency contact using the continuity note.

5. Testing

  • Annual: one VM-rebuild rehearsal (SOP-06 §4) and one backup-restore drill.
  • The result, duration and lessons are logged in the risk register (POL-11).
← All documents & policiesQuestions? hello@jonnyai.co.uk