Managed infrastructure

Small hardware. Serious recovery plan.

A production backend needs more than a computer. It needs power behavior, backups, secure connectivity, monitoring and a tested path back from failure.

Reference deployment

The standard small-business server stack.

These exact models are reference hardware—not fake renders. Final hardware is selected against workload, availability and client requirements at the time of purchase.

BEELINK · EQ14 N150

16GB DDR4 / 500GB SSD

Compact x86 server reference for Docker services, application backend, relational database, automation and Cloudflare Tunnel.

Server

SAMSUNG · MU-PE2T0S/AM

T7 Shield 2TB

Dedicated local encrypted backup target with enough room for versioned database, configuration and file snapshots.

Backup

APC · BX850M

850VA / 510W UPS

Battery backup with AVR and USB data interface for short-outage continuity and a controlled shutdown during longer outages.

Power

GL.iNET · GL-X3000

Spitz AX 5G Multi-WAN

Optional cellular gateway for locations that need an alternate internet path and automatic failover when the primary WAN is unavailable.

Optional

Product references are maintained by exact model so a proposal can distinguish recommended architecture from whatever happens to be cheapest on a marketplace that day.

Power-loss behavior

What should happen when the building goes dark.

The goal is predictable behavior, not hope. Short outages stay online; long outages shut down cleanly; restored power brings the stack back automatically.

Utility fails

UPS immediately carries the mini PC and essential network equipment.

Battery monitored

The server reads UPS state rather than waiting for an abrupt battery cutoff.

Graceful shutdown

Database and services stop cleanly when the configured battery threshold is reached.

Power returns

BIOS power-restore behavior boots the mini PC and service manager starts the stack.

Health verified

Tunnel, database, application and backup status are checked and reported.

3-2-1 thinking

Backups are useful only if they can be restored.

The baseline design keeps the live data on primary storage, a versioned local backup on separate physical media, and an encrypted off-site copy outside the building. Periodic restore testing proves the recovery path.

Frequent database snapshotsProtect recent operational changes.

Versioned retentionRecover from a mistake discovered days later.

Off-site encryptionSurvive theft, fire or local hardware loss.

Documented rebuildProvision a replacement machine without cloning a dead one.

Optional continuity upgrade: add cellular failover when technicians or remote users still need the backend during an ISP outage.

Higher availability: when an hour of backend downtime is unacceptable, add a standby system or cloud replica rather than pretending one mini PC is enterprise HA.