Home › Topics › The IT Department of One › Solo Setup

The Solo Admin's Monitoring Setup: Maximum Coverage, Minimum Upkeep


Most monitoring advice quietly assumes a team. Someone to share the pager. Someone to glance at the dashboard while you are at lunch. Someone to ask "hey, did you see that alert?" When the whole team is you, none of that exists. So every design choice in your monitoring setup has to pass through one filter first: what does this cost me in ongoing attention? Suppose a check needs weekly babysitting. An alert cries wolf twice a month. A dashboard only works if somebody watches it. On a team, these are minor annoyances. For a department of one, they are debt collected from the only account you have.


The good news: a monitoring setup designed for one person is not a smaller, sadder version of a team's setup. It is a different design, and in some ways a better one, because the constraint forces clarity. This guide walks through the solo design principles and the architecture choices that keep the system boring. It gives you a three-lane alert routing plan that protects your attention, and a complete worked example for an 80-device company. It ends with a target you can hold yourself to: under two hours of monitoring upkeep per week.


In this guide

  • The attention budget: the question that designs everything
  • The four solo design principles
  • An architecture that mostly runs itself
  • The three-lane routing plan
  • The failure mode your monitoring cannot fix
  • A worked example: 80 devices, one admin
  • The solo setup checklist
  • Where to go next

The attention budget: the question that designs everything


Think of your attention as a budget, the same way you think of disk space or bandwidth. Every check you create, every alert rule you write, every report you schedule either spends from that budget or saves into it. It spends by demanding your eyes and your interruptions. It saves by noticing things so you do not have to. A team can afford some waste, because the cost is spread across several people. You cannot. An alert that interrupts you needlessly does not just cost the two minutes it takes to dismiss. It costs the focus you were using on whatever it interrupted, and a small piece of your trust in the system.


So before any tooling decision, ask the filter question of every element: if I add this, what does it cost me per week, forever? A check that fires only when something is truly wrong costs almost nothing and pays out constantly. A check that needs its threshold adjusted every month is a part-time employee you have to manage. The whole rest of this guide is that one question, applied systematically.


The four solo design principles


Principle 1: fewer, better checks

Your first instinct might be to monitor everything, since nobody else is watching. Resist it. The essential set includes internet link, firewall, core switch, DNS, DHCP, the servers everyone touches, backups, power, Wi-Fi, and certificates. That set covers the vast majority of real risk on a small network. It is small enough that every alert still means something. The guide to the first ten checks walks through exactly this set with starting thresholds. For a solo admin, that set is not just a starting point; it is close to the whole strategy. Add checks beyond it only when each one passes the test: if this fired tomorrow at 9 AM, would I act? If the honest answer is no, you have not added coverage. You have added noise with a delay timer.


Principle 2: defaults that age well

Hand-crafted, per-device check configurations are a luxury for teams with time. You want templates: one standard definition for "a Windows server," one for "a switch," one for "an access point." Each definition is applied to every device of that kind. When you learn something — say, that your disk warning threshold is too twitchy — you fix the template once. Every device inherits that fix. Hand-tuned snowflake checks drift apart silently, and six months later you cannot remember why server three alerts at 80% and server four at 90%. Templates are also self-documenting: a new device gets the standard treatment automatically. That matters when the person applying the standard is also the person who forgot writing it.


Principle 3: aggressive noise discipline from day one

On a team, a noisy alert annoys five people a little. Solo, all of it lands on you, and there is no one to compare notes with when you start tuning it out. That is how alert fatigue kills monitoring systems, and solo admins are its easiest victims. So make quietness a default, not a cleanup project. Put confirmation and hysteresis on everything where they make sense. Confirmation means the condition must be seen several checks in a row before alerting. Hysteresis means the alert clears at a better value than the one that triggered it, so it cannot flap on the boundary. The guide to hysteresis and confirmation covers the mechanics. The solo rule is simpler: any alert that has interrupted you twice without requiring action gets tuned or demoted that same day. You do not have the headcount to be patient with noise.


Principle 4: automation that tends itself

Two features earn their keep more than any others in a solo shop. The first is auto-discovery: the monitoring system scans your network on a schedule and tells you what appeared or vanished. That keeps your inventory honest without you maintaining a spreadsheet — and on a solo network, the spreadsheet was never going to stay maintained. The second is scheduled reports: a summary that arrives on its own replaces a dashboard you would have to remember to look at. The difference sounds cosmetic but is structural. A report is push, a dashboard is pull, and pull systems fail quietly the week you get busy. Which, solo, is every week.


An architecture that mostly runs itself


One modest, boring monitor

You do not need a monitoring cluster, a message queue, or anything with the word "distributed" in it. You need one modest, reliable machine — a small VM or a spare box — placed somewhere boring and stable. Put it on the main network segment it watches, on protected power. Do not put it on a host it is responsible for monitoring (or one crash takes out the patient and the doctor together). Resist any architecture you could not explain to your future self on a bad day. Complexity in the monitor is upkeep you will pay for monthly; a single boring server is a design decision, not a compromise.


Backups and the afternoon rebuild test

The monitoring server must be recoverable without heroics, because the person doing the recovering will be you, mid-outage, annoyed. Back up its configuration automatically — most monitoring tools can export or back up their config on a schedule. Keep a short rebuild note where you keep documentation. The standard to hit: you could rebuild the whole thing in an afternoon from your notes. If you cannot say that, simplify until you can. A rebuild note fits on one page:


MONITOR REBUILD NOTES - keep with the IT documentation
1. VM spec: 2 vCPU / 4 GB RAM / 60 GB disk. Name MON01, IP 10.0.0.25
2. Install monitoring software; restore config backup from \\nas\it\monitor-config\
3. Credentials (SNMP community, WMI service account): password vault, "Monitoring" folder
4. Discovery ranges: 10.0.0.0/24 (servers/net gear), 10.0.20.0/24 (APs/printers)
5. Alert routing: three lanes - see routing table in this document
6. Before trusting it: force one test alert per lane and confirm delivery

The calendar as co-admin

On a team, recurring maintenance survives because someone eventually says "isn't it about time we tested a restore?" Solo, nobody says it, so the calendar has to. Put the recurring tasks in your actual work calendar — the one with meeting reminders that you cannot ignore. Make them real appointments, not a list in a document. The quarterly set that keeps a solo monitoring setup healthy:


Recurring taskCadenceWhat it involves
Threshold reviewQuarterlySkim alert history; tune or demote anything that fired without needing action
Restore testQuarterlyRestore one real file and one config from backup; note how long it took
Certificate and renewal sweepQuarterlyConfirm cert checks cover every service; check domain and support renewals
Monitor self-careQuarterlyUpdate the monitoring server, verify its config backup, reread the rebuild note
Discovery reviewMonthlyReview what auto-discovery found or lost; add or retire checks accordingly

Ninety minutes a quarter, scheduled like meetings with a colleague who happens to be a calendar. It is the cheapest reliability you will ever buy.


The three-lane routing plan


Here is where the attention budget becomes concrete. Every alert your system can produce gets sorted into exactly one of three lanes, by one criterion: how soon does this genuinely need a human? Not how important the device is, not how scary the metric sounds — how soon action is needed.


LaneDeliveryBelongs hereExamples
Page-worthyPhone notification, immediately, any hour it appliesSomeone must act now or the business visibly suffersInternet down, firewall or core switch down, virtualization host down, UPS on battery, server room hot
Morning digestOne email, ready when you start the dayNeeds action soon, but tomorrow morning is soon enoughBackup job failed, disk at 85%, one AP down, cert inside 30 days, DHCP scope at 80%
Weekly summaryOne report, same time every weekTrends and drift; no single event needs actionCapacity growth, error-counter trends, newly discovered devices, uptime summary

The discipline is in what the page lane excludes. A failed backup at 2 AM is a genuine problem — and also a problem that is exactly as fixable at 8 AM. So it rides the digest. Solo, the page lane is not "important things"; it is "things worth waking the entire IT department for." The entire IT department needs to be sharp tomorrow. Deciding lane by lane forces the honest question a team never quite has to answer: is this worth interrupting the only responder we have?


The morning ten minutes

The digest enables the single best habit in solo operations: a fixed ten-minute morning ritual. Coffee, digest, monitoring overview, done. You are looking for three things: anything that fired overnight, anything trending toward a threshold, and anything that looks different from yesterday. Ten focused minutes replaces the ambient anxiety of checking dashboards all day — and ambient anxiety, not workload, is what actually wears solo admins down. You are allowed to not think about monitoring for the other 470 minutes of the workday. That is the system working.


What not to build

Skip anything that exists to be watched: the wall-mounted dashboard TV, the real-time graphs styled for a network operations center, per-user bandwidth leaderboards. Those are team furniture — they assume an audience, and your audience does not exist. (There is one exception. A simple status view that a non-IT colleague can read makes sense as part of your vacation coverage. That is a different job with its own guide.) Every element of a solo setup should either interrupt you correctly or summarize for you compactly. Anything else is decoration that will one day need upgrading.


Tip: when you build the three lanes, force a test alert through each one. Unplug something safe, or set a threshold to an absurd value for a minute. You want to learn how each lane feels on your phone and in your inbox before a real outage teaches you. Do this again whenever notification settings change; a routing rule that silently broke is worse than none.


The failure mode your monitoring cannot fix


An honest section, because the machinery above has one gap it cannot close: you are the single point of failure. The monitoring can detect every problem on the network and it still cannot fix one fact. Only one human can interpret the alerts, hold the credentials, and know that the finance server is fragile about reboots. Flu, jury duty, a resignation someday — the risk is not exotic.


You cannot hire your way out of this today, but you can shrink it steadily. Write down what only you know, starting with the checks you built and why. The reasoning behind a threshold is exactly the kind of knowledge that lives only in your head. If your network is currently one of those where the documentation is you, the undocumented network guides cover digging out of that hole. The full treatment of making yourself temporarily replaceable — coverage plans, the break-glass document, delegate briefings — is the vacation guide's territory. That treatment doubles as your bus-factor insurance: Taking a Real Vacation as a Department of One.


A worked example: 80 devices, one admin


To make this concrete, here is a complete solo setup for a realistic small company. The environment has one office, one internet connection, a firewall, and a core switch plus five access switches. It has two virtualization hosts running a dozen VMs (domain controller, file server, application server, print server, backup server among them). It has a NAS receiving backups, two UPS units, eight access points, six printers, and about fifty workstations. Cloud email. One admin: you.


The check list — note how much is deliberately left out:


WhatChecksLane on failure
Internet linkPing external targets; latency and lossPage
FirewallReachability, CPU, memory, WAN portPage
Core switchReachability, uplink ports, port errorsPage (down) / weekly (errors)
Access switches (5)Reachability, uplink statusMorning digest
Virtualization hosts (2)Reachability, CPU, memory, datastore spacePage
Key VMs (6 of 12)Up, CPU, memory, disk, key service runningPage (DC, file) / digest (rest)
DNS and DHCPReal lookups; service up; scope usagePage (both DNS down) / digest (scope)
BackupsJob success reports; missing report = failureMorning digest
UPS units (2)On battery, battery health, runtime, temperaturePage
Access points (8)Up; client countsMorning digest
Certificates and domainsDays to expiry, dailyMorning digest
Cloud email reachabilityCan we reach it from here, and how fastMorning digest
Workstations, printers, test VMsNot monitored. They generate tickets, not alerts—

That is roughly forty to fifty individual checks from about thirty monitored objects, nearly all of it applied through three or four templates. The page lane contains perhaps ten possible alerts, each of which would genuinely justify standing up from dinner. Everything else arrives in one morning email and one weekly report. The weekly time budget:


ActivityTime
Morning ritual (10 min × 5 days)50 min
Weekly summary review (Friday)20 min
Tuning, gardening, discovery follow-ups20 min
Quarterly calendar tasks, averaged weekly10 min
Total~100 minutes

Under two hours a week, for coverage of everything that can take the company down. If your setup is costing you more than that, the overage is almost always in one of two places. One is alerts that fire without needing action (tune them ruthlessly). The other is manual work the system should be doing for you (discovery, reports, config backup). The budget is not a stretch goal — it is what a correctly designed solo setup naturally costs.


The solo setup checklist


Everything above, as a list you can work through and check off:


#ItemDone when
1Essential checks deployedThe first-ten set is live; nothing beyond it without a "would I act?" yes
2Templates, not snowflakesStandard device types get standard checks; tuning edits templates
3Confirmation and hysteresis onNo alert fires from one bad poll; none flaps on a boundary
4Auto-discovery scheduledNew and vanished devices appear in a report, not by surprise
5Three lanes routedEvery alert assigned page / digest / weekly; test alert sent through each
6Morning digest arrivingOne email covers overnight events and trends before 8 AM
7Monitor placed and protectedBoring host, protected power, not on hardware it monitors
8Config backed up + rebuild noteYou could rebuild in an afternoon; the note says how
9Calendar co-admin activeQuarterly tasks exist as real calendar appointments
10Noise rule enforcedTwice-interrupted-without-action alerts get tuned the same day

Where to go next


A setup like this does something subtle beyond catching outages. It converts monitoring from a source of ambient worry into a bounded, scheduled activity. That means ten minutes in the morning, a report on Friday, a quiet phone the rest of the time. That reclaimed attention is the raw material for everything else in solo IT. That includes the automation that buys back more hours, the sustainable pace that keeps you sharp, and eventually the evidence that gets you help. The rest of The IT Department of One picks up each of those threads.


Put this into practice with SysPeep Monitor

A solo-friendly setup is mostly about what the software does unattended. SysPeep Monitor auto-discovers the devices and interfaces on your network and watches them over SNMP, WMI and SSH. Its thresholds, dependencies and alert rules are built to keep the noise down. So the page lane stays quiet and the summary lands in your inbox. It runs on one Windows machine, and the free trial takes an afternoon, not a project plan.

Download Free TrialLearn About SysPeep