Why Your Website Needs Uptime Monitoring
You monitor a website so that you learn it is broken before your customers do, instead of after. Without a monitor, the first report of an outage usually arrives from a visitor on social media, by email or by phone, and by then the outage has been running for however long it took someone to bother telling you.
What takes a site down
A site can stop answering for reasons that have nothing to do with each other. The usual list:
- Bad code, including a deployment that passed review and failed in production
- Web host failures, from a single machine to a whole data center
- Host limits reached: bandwidth, connections, CPU quota, disk
- Hacking attempts against the application or the server
- DDoS attacks, which overwhelm the host until it stops answering any request, good or bad
- An expired domain registration
- An expired TLS certificate, which makes every browser refuse the connection
Most of those are within your control. You can review code, size the hosting to the traffic, and renew a domain and a certificate on time. Hacking attempts and DDoS traffic are not: you can only limit and absorb them. What all seven have in common is that the server rarely tells you. A machine that cannot reach the internet cannot send you a message saying so, and a certificate that expired at midnight does not file a ticket.
What an outage costs while you are unaware of it
A visitor who hits "Server Not Found" refreshes once, checks that the address was typed correctly, and then looks for a site that works. The window in which you keep that person is short, and it closes without anyone telling you it did.
The arithmetic is easy to run for your own site. Take an average order value of $25 and 50 visitors an hour who convert: that hour is worth about $1,250, and an hour of downtime costs the same $1,250. Nothing in your analytics records that hour as a loss: the orders that did not happen leave only a dip you will attribute to something else next week. Slow pages do a quieter version of the same damage. The site answers, the check passes, and visitors leave anyway.
How monitoring changes the sequence
Monitoring replaces "a customer told us" with "the check failed at 03:14 from four locations".
- It starts the clock earlier. Time to repair is measured from the moment you know, so anything that shortens the discovery gap shortens the outage.
- It hands you evidence. Which check failed, from which locations, with which status code or timeout, and when the last good result was. That is most of a diagnosis before you have opened a terminal.
- It tells you the difference between down for everyone and down from one network, because the checks run from many places at once rather than from a single server.
External checking is the part a server-side log cannot replace. If the problem is the network path, the DNS record or the certificate, the server sees nothing wrong at all. Running the same request from many countries, which is the idea behind distributed availability monitoring, is what makes a regional failure visible as a regional failure rather than as a mystery.
What to monitor beyond up or down
An availability check answers one question. The things that break a site tend to be visible in other places first:
- Server uptime and downtime, the baseline check
- Response time, which degrades before it fails
- CPU, RAM and disk pressure on the server
- Database load and query latency
- Domain expiration
- Certificate expiration
- Page content, so a site that returns 200 with an error page is still caught
Domain and certificate expiry deserve a note of their own, because they are the two outages with a known date in advance. Both are entirely preventable by an alert set weeks ahead.
Reducing downtime once you can see it
- Put a monitor on the site, from outside, on an interval short enough that an outage is measured in minutes rather than hours. Run a one-off HTTP check from multiple locations first to see how the site responds today.
- Choose a host with capacity headroom and a record you have checked rather than one you were told about. An "unlimited" plan is not unlimited.
- Keep the application, the server software and the plugins patched. A compromised site loses more than availability: it loses standing with customers and with search engines.
- Set expiry alerts for the domain and every certificate, with enough lead time to renew.
Checking once versus knowing when it breaks
A check you run by hand answers for the moment you ran it. A monitor answers continuously and tells you when the answer changes. HostTracker has monitored websites since 2004 and now watches over 500,000 websites from 300+ checkpoints in 158 cities, across 13 monitor types, with alerts by email, SMS, voice call, Slack, Telegram and more. See the feature set for what those checks cover, the HTTP check guide for the one most sites start with, and the rest of the monitoring concepts section for the terms behind them.