Connectivity strategy — Claxton home base and Demory farm site¶

Purpose¶

Give a single canonical statement of how the two-site farm expects to reach the Internet and administer remote systems, with named places: SITE_HOME — Claxton / Anderson County; SITE_FARM — Demory / Campbell County. This page does not replace the diagram-level Starlink/WAN analysis (Two-site smart farm — network topology and WAN/edge reference (Mermaid)) or the logical stack (Reference architecture — 5 ac + 120 ac)—it routes decisions and names assumptions for pilots and scale-up.

Execution topology package (reference / pilot / degraded Mermaid): Execution topology package — two-site smart farm (Mermaid).

First-24-months validation track (survey, power, WAN testing, remote access, cost stops): Validation and pilot plan § Connectivity validation.

Supplementary diagrams (telemetry plane, DC, HaLow overlay): Two-site smart farm — network topology and WAN/edge reference (Mermaid).

Raw / spec grounding (PDFs and captures, not operational truth at coordinates): Electrical, networking, and Starlink — inbox batch (2026-04-23).

SITE_FARM / Demory — off-grid-first planning stance: solar + battery default; local mesh / field RF before relying on Starlink / LTE for operations—Demory farm — site intelligence § Off-grid-first, Off-grid power strategy — Demory farm site, Mesh and field networking strategy — off-grid Demory farm.

Field sensor plane (LoRaWAN / HaLow / Meshtastic; Starlink is not field RF): Farm sensor architecture — Demory farm site.

Starlink / LEO broadband — role by site¶

Enables (both sites, when deployed): batched trips with remote checks, admin/VPN paths, MQTT/cloud egress from field gateways, maps and comms while mobile.

Must not silently become: the only path for authoritative books without offline export; a substitute for physical water/livestock checks; assumed bandwidth for dense camera mesh; proof of coverage at coordinates (specs ≠ field—validate per Validation and pilot plan — first 24 months).

Terrestrial Internet assumptions¶

• Homestead / Claxton: Cable or fiber may exist—survey and log addresses before locking CAPEX. If wireline is absent or poor, LEO satellite (e.g. Starlink-class) is a realistic first stable Internet for the control center; still document latency, data caps (if any), and outage behavior.
• Farm / Demory: Buried fiber to the field is not assumed day one. Last mile may be LTE, satellite, or none—field gateways should buffer and fail safe when WAN is gone (Manual fallback and degraded modes).

LTE / cellular fallback assumptions¶

• LTE is the usual backup or parallel path where tower coverage exists—drive-test and log bars vs usable data at pins that matter (gates, tanks, barn), not only at the driveway (Validation and pilot plan V10 angle).
• Assume carrier CGNAT, variable routing, and possible throttling—do not depend on stable inbound to field CPE for admin without outbound-initiated tunnels or a rendezvous you control (Remote access and operational security model).
• Satellite WAN impaired (rain fade, obstruction, power loss at terminal) should fall through to LTE where provisioned—otherwise treat as comms down for remote trust (Automation degraded modes SOP failure class A2).

Local-only versus WAN-dependent systems¶

Architecture boundary: Reference architecture — 5 ac + 120 ac — telemetry vs SoR, broker placement.

Power dependencies and outage implications¶

• Starlink (and most CPE) needs continuous power: WAN loss during utility outage may be even if radios still work locally—do not equate “generator on” with “Internet up” without explicit UPS/branch design for modem/terminal (see DC / AC patterns in Network topology (Mermaid) and power entities Farm on-site power system).
• LTE routers share the same rule: no power → no fallback path.
• Local mesh / LoRa may outlive WAN if gateways and sensors stay powered—alerts may silence without implying “all clear” at the animal or water level.

Cost implications¶

• OPEX: treat ISP + cellular + LEO subscription as explicit line items (not generic “tech”)—see Capital plan and infrastructure sequencing Tech row and Starlink / farm WAN sequencing note.
• Deferral discipline: second terminal or redundant farm WAN after Phase 1 comms proof—align with Two-site operations model — 5 ac / 120 ac WAN table.
• Spares: budget cable, router, or radio spares proportionally to revenue risk of silent telemetry—not zero, not infinite.

Remote access and security implications¶

• LEO path is still untrusted Internet (Z3a in the security model)—encrypted tunnels, scoped creds, no flat admin from satellite “privacy” feel (Remote access and operational security model).
• CGNAT / dynamic addressing favors outbound VPN or controlled rendezvous—document one primary admin pattern.

Network policy (segmentation of cameras, gateways, telemetry, admin, and user devices; site-to-site assumptions; VPN/overlay; what may be Internet-reachable): Network segmentation, site-to-site connectivity, and internet exposure — two-site smart farm. MV/CS hooks: Remote access… § Execution and validation hooks.

Farm functions: WAN required vs must tolerate WAN loss¶

Pilot phase vs scale phase — connectivity expectations¶

Gates and calendar: Validation and pilot plan — first 24 months; phase bands: Phase timeline.

Related pages (execution spine)¶

Operations hub¶

Router: Two-site smart farm operations.