Monitoring and Alerting: Knowing Something Broke Before Your Customers Do

Monitoring sounds like an enterprise concern, something for companies with a network operations centre and a wall of screens. It isn't. At its core it's a simple, universal need: to find out that something important has broken before your customers, suppliers, or your own staff do. For a small business, getting a quiet text message that your site is down at 2pm is worth a great deal more than discovering it from a lost sale at 5pm.

The distinction that matters most: monitoring vs alerting

These two words get used interchangeably, but they're different jobs, and confusing them is why so much monitoring is useless:

• Monitoring is collecting information about whether things are working: is the site responding, is the disk filling up, did the backup run.
• Alerting is being told, actively, when something's wrong. It's a message that reaches you, now, without you having to go and look.

What's actually worth monitoring

You can monitor almost anything, which is exactly why people monitor the wrong things and drown. For a typical small business, this short list covers the overwhelming majority of "oh no" moments:

Is it up? (Availability)

The most basic and most valuable check: is your website responding, is your email server accepting mail, are the services customers rely on actually reachable? This catches the outages that directly cost you money and reputation. It's also the easiest to set up and the one with the highest payoff.

Is the certificate about to expire? (SSL)

An expired SSL certificate takes a site offline just as effectively as a crashed server, the browser blocks it with a scary warning. With certificate lifespans shrinking, expiry is an increasingly common, entirely preventable outage. A monitor that warns you days before a certificate expires turns a crisis into a non-event.

Is it running out of room? (Disk and resources)

A disk that fills up will take down a website, a database, or a mail server, and it does so gradually, then all at once. Watching free disk space (and, to a lesser degree, memory and load) catches a slow-building problem while there's still time to act, rather than at the moment everything stops.

Did the backup actually succeed? (Backups)

As covered in our backup guide, a backup you assume is running but never verify is one of the most dangerous false comforts in IT. Monitoring that your backups are actually completing, and alerting you when one fails, is the difference between "we're covered" and finding out otherwise on the worst possible day.

The principles that keep it useful

Good monitoring is less about tools and more about discipline. A few principles do most of the work:

• An alert must be actionable. Every alert should mean "a human needs to do something." If an alert doesn't require action, it shouldn't be an alert, it's noise, and noise is actively harmful.
• Beware alert fatigue. This is the single biggest killer of monitoring. If you get twenty alerts a day that don't matter, you'll start ignoring all of them, including the one that does. Fewer, higher-quality alerts beat comprehensive noise every time.
• Monitor from outside, not just inside. A server checking its own health can't tell you it's unreachable: if it's down, so is its own monitoring. An external check (from another location entirely) is what catches a total outage. Ideally you have both.
• Make the alert reach you where you'll see it. An email alert is useless if the outage is your email. Route critical alerts through a channel independent of the thing being monitored, a text message, a push notification, a separate service.
• Test that the alerting actually works. Like backups, an untested alert is a hope. Deliberately trigger a test and confirm the message actually reaches you, before you're relying on it for real.

Keeping it proportionate

You don't need an elaborate monitoring stack to get most of the benefit. There are excellent, approachable tools, including free and open-source ones, that will watch whether your key services are up and message you when they're not, and that alone puts you far ahead of most small businesses. Start with availability and certificate monitoring for the things customers touch, add backup-success and disk-space checks for the things that fail quietly, and make sure every alert reaches you somewhere you'll actually notice. That's a proportionate setup that earns its keep the first time it saves you from finding out the hard way.

Frequently asked questions