Reference architecture — 5-acre home base + 120-acre farm¶

Purpose¶

Give a single first-draft logical picture of the smart-farm stack for the staged East Tennessee two-site business plan: SITE_HOME_5AC as control center, workshop, smart home, and data/telemetry hub; SITE_FARM_120AC as production and remote operations target. Not a bill of materials—boundaries, paths, and failure domains.

Business plan links: Planning framework, Smart technology and telemetry strategy, Two-site operations model — 5 ac / 120 ac. Connectivity (WAN) strategy (Claxton / Demory): Connectivity strategy — Claxton home base and Demory farm site.

Entity anchors (modeled sites and systems—no parcel-specific claims): Five-acre home base (SITE_HOME), 120-acre production farm (SITE_FARM), Field telemetry network — two-site.

Package map (this wiki)¶

#	Deliverable	Page
1	Reference architecture (this page)	Mermaid execution views (reference / pilot / degraded): `Execution topology package — two-site smart farm`
2	Telemetry system of record — boundaries & authority	`Telemetry system of record — boundaries and authority`
3	Asset registry & spatial model — minimum standard	`Farm spatial model and asset registry standard`
4	Instrumentation priority matrix	`Instrumentation priority matrix`
5	Automation principles (early / late / never)	`Automation principles — two-site smart farm`
6	Degraded modes & manual fallback	`Manual fallback and degraded modes — critical operations` + `Automation degraded modes SOP`
7	Remote access & operational security	`Remote access and operational security model`
8	Network segmentation, site-to-site policy, internet exposure	`Network segmentation, site-to-site connectivity, and internet exposure`
9	Off-grid `SITE_FARM` (Demory) — power + field RF + optional WAN	`Off-grid systems doctrine package — Demory` · `Off-grid power doctrine — Demory farm site` · `Mesh and field networking strategy — off-grid Demory farm` · `Off-grid farm execution topology — Demory (Mermaid)`

Deeper technical stack (MQTT, broker, radio paths): Field telemetry reference architecture — homestead + 120-acre farm. Mermaid topology (WAN, sites, telemetry plane, DC context): Two-site smart farm — network topology and WAN/edge reference (Mermaid). Off-grid Demory (solar+battery default, mesh-first): Off-grid farm execution topology — Demory (Mermaid).

Homelab backup / DR (farmOS-class DB, containers, CSI volumes): Backup and disaster recovery package — smart farm stack · Backup strategy comparison — farmOS, homelab, PostgreSQL, containers · Restore and recovery tiers — homelab farm systems · Runbook — backup validation and recovery drill · Kubernetes platform backup / DR — Pi, k3s, Longhorn · Off-grid implications — backup and networking choices.

Homelab Kubernetes platform (Pi fleet, k3s, Longhorn, optional Rancher): Platform doctrine package — homelab / farm edge · Platform strategy for farm and homestead services · Platform decision memo · Provisioning runbook: How to provision k3s, Longhorn, and Rancher on a Raspberry Pi fleet.

Placeholders (edit as you deploy)¶

Symbol	Meaning
`COMMUTE_ONE_WAY`	Typical drive time home → farm (minutes)—drives batching and acceptable outage windows
`SITE_HOME_5AC`	Control center: LAN, NOC, workshop tooling, cold chain option, primary internet (often LEO satellite broadband such as Starlink if wireline absent—see WAN note below)
`SITE_FARM_120AC`	Production: paddocks, water, livestock/plant systems, field gateways, weaker or no wireline (Starlink or LTE uplink when justified—not assumed day one)

Starlink / WAN (explicit placement)¶

Diagrams: Two-site smart farm — network topology and WAN/edge reference (Mermaid).

Site	Starlink-style LEO WAN	Architecture note
`SITE_HOME_5AC`	Primary or backup relative to any future fiber/cable	Edge router + policy sits after the terminal; homelab does not assume flat L2 to the 120 ac—VPN/overlay reunifies logic, not broadcast domains.
`SITE_FARM_120AC`	Conditional / deferred	Field gateways need some uplink for cloud/MQTT; whether that is second terminal, LTE, or shared policy is a Phase 1 decision—avoid funding WAN duplication before water/fence gates.

Enables: remote observability and admin paths consistent with Remote access and operational security model.

Must not silently become: proof of coverage at coordinates (spec PDFs ≠ field); substitute for safe physical defaults at SITE_FARM_120AC when backhaul drops (Manual fallback).

What changed because of Starlink analysis: The WAN box in the Mermaid page is labeled Starlink as an example LEO path—this reference page inherits that placement without repeating vendor prose.

Layered view (logical)¶

Layer	Primary location	Role
Human ops	Both; brain often at `SITE_HOME_5AC`	Decisions, manual verification, sales/logistics
Records / books	Usually `SITE_HOME_5AC` (office)	Authoritative for financial and compliance narrative—see SoR page
Telemetry ingest	`SITE_FARM_120AC` edges → backhaul → hub	Measurements; not automatically legal records
Broker / message bus	Hub or cloud—decision	Transport; ephemeral unless persisted per SoR rules
History / DB / TSDB	Hub or cloud	Durable sensor history when configured
Dashboards / HA	`SITE_HOME_5AC`	Convenience and automation; can lie if broker stale
Alerts	Phone / email / SMS	Attention routing—not a database

Authority rules (detail): Telemetry system of record — boundaries and authority.

Video / NVR (explicit boundaries): IP camera stacks (e.g. Frigate) are optional and not welfare-grade proof alone; SITE_FARM production dependence is deferred per Local video / NVR — role and deferral boundaries (farm stack). Observability + secrets doctrine: Observability, secrets, and trust bar — homelab / farm edge.

Data and control flows (text diagram)¶

text [ Operators @ SITE_HOME_5AC ] <--- commute ---> [ Field work @ SITE_FARM_120AC ] | | | dashboards / alerts | sensors / actuators / gateways v v [ Hub: broker + persistence + automation core ] <---- backhaul (cell / fiber / PtP / mesh TBD) ^ | optional VPN / tunnel | [ Internet / DNS / remote admin paths ] ---> see security model page

Starlink note: Replace generic “cell / fiber / PtP / mesh TBD” mentally with documented choices: LEO satellite at homestead is a common rural primary; farm uplink may lag homestead or use LTE—see Two-site operations model — 5 ac / 120 ac WAN table.

What “control center” must cover (minimum)¶

Function	Why at 5 ac
Stable internet (or planned failover)	Remote view of 120 ac dies otherwise
Workshop + spares	Field gateway swap without city detour
Documentation	Runbooks live somewhere durable—not only in heads
Secure admin surface	Patching without exposing whole farm LAN
Optional cold / office	Logistics and food safety paperwork

Scenarios (must be designed, not assumed away)¶

Scenario	Design pressure
`COMMUTE_ONE_WAY` too long for emergency	Tier risks; neighbor / hired coverage; reduce need for instant remote fix
Backhaul down	Local logging or safe defaults; increase physical checks—`Manual fallback`
Broker / cloud down	Same as above; no silent confidence in dashboards
Sensor wrong	Disable closed-loop on that input; manual measurement—`Sensor triage runbook`
Maintenance backlog	Fewer sensor types beat more half-maintained devices—`Automation principles`

Open items¶

One-page network list: subnets, VPN, broker hostname, who can admin from where.
Map device_id → paddock_id for first N instruments—Asset registry.
Tabletop exercise: Friday night storm + LTE down + water unknown.