What is uptime? What 99.9% really means, and how to monitor it

What website uptime is, what 99%, 99.9%, 99.99% and 99.999% mean in minutes of downtime, how SLAs work, and how to monitor your own site.

1. Uptime in one paragraph

Uptime is the share of time a website or service is up and working, usually given as a percentage over a month or a year. 99.9% uptime allows about 43 minutes of downtime a month; 99.99% allows about 4 minutes. Each extra nine cuts the allowed downtime by ten, and is much harder and more expensive to achieve.

2. Uptime percentages in minutes of downtime

Percentages hide how much downtime they allow. Here's what each common figure means, using a 30-day month and a 365-day year:

• 99% (two nines): about 7 hours 12 minutes a month; about 3 days 15.6 hours a year.
• 99.5%: about 3 hours 36 minutes a month; about 1 day 19.8 hours a year.
• 99.9% (three nines): about 43 minutes a month; about 8 hours 46 minutes a year.
• 99.95%: about 22 minutes a month; about 4 hours 23 minutes a year.
• 99.99% (four nines): about 4 minutes 19 seconds a month; about 53 minutes a year.
• 99.999% (five nines): about 26 seconds a month; about 5 minutes 15 seconds a year.

To work it out yourself, multiply the length of the period by the share of downtime allowed. For 99.9% over a 30-day month:

30 days x 24 hours x 60 minutes = 43,200 minutes
Downtime allowed: 100% - 99.9% = 0.1%
43,200 x 0.001 = 43.2 minutes

A 31-day month allows a little more, and many providers measure per calendar month, so check how they count.

3. What counts as downtime

It sounds simple, but providers and tools define it differently:

• Completely unreachable: the site doesn't answer at all. Everyone counts this.
• Errors: the site answers with an error such as 500, 502 or 503. Most monitors count this as down.
• Too slow: a page that takes a minute to load is down for your visitors, but many uptime figures don't count it.
• Partly down: the home page loads but checkout or login fails. A simple check of one page misses this.
• Down in one region: visitors in one country can't reach you while others can.
• Planned maintenance: often excluded from SLA figures, even though visitors can't use the site.

When you compare uptime claims, ask what's measured, from where, how often, and what's left out.

4. Uptime guarantees and SLAs

A service level agreement (SLA) is a written promise about uptime, usually with a remedy if it's missed. The remedy is almost always a service credit: part of that month's fee back, often only if you claim it. That's rarely anything close to what an outage costs a business.

So treat an SLA as a signal of how seriously a provider takes reliability, not as insurance. Read the exclusions: scheduled maintenance, problems caused by your own code or settings, attacks and outside services are commonly left out.

A host's uptime is also only part of yours. Your site can be down while the host is perfectly healthy.

5. Why websites go down

On the host's side: hardware failures, network problems, overloaded servers, maintenance and attacks. On your side, which is more common than people expect:

• An expired domain. If it isn't renewed, your site and email stop. Turn on auto-renew and keep your payment details current.
• An expired HTTPS certificate. Browsers then show a full-page warning. Good hosts renew certificates automatically.
• A broken deploy. A new version that crashes on start. Hosts with health checks keep the working version running until the new one passes.
• A database or outside service failing. Your site is up, but every page that needs the database shows an error.
• Running out of resources. Memory, disk or traffic limits. Some hosts take a site offline when it goes over its plan.
• DNS mistakes. A changed or deleted record can send visitors nowhere.

6. How to monitor your site's uptime

Don't rely on your host or your visitors to tell you. An uptime monitor checks your site from outside, the way a visitor would, and alerts you when it fails. Many monitoring services have free plans.

1. Choose what to check. Your home page, plus one page that uses your database, or a health endpoint such as /api/health that tests the important parts.
2. Set what success looks like. Usually a 200 status code. For more confidence, check the page contains a word that should always be there.
3. Check often. Every minute catches short outages. Every five minutes or more can miss them.
4. Confirm before alerting. Retry once, ideally from a second location, before calling the site down. That cuts false alarms.
5. Send alerts where you'll see them: email, a text or a team chat. Test that they arrive.
6. Watch the edges: certificate expiry, domain expiry, and scheduled jobs (with a heartbeat check that alerts when a job doesn't report in).

Over time the monitor gives you your own uptime figure, measured your way. That's more useful than any number on a pricing page.

7. How to improve your uptime

• Choose a host whose sites don't sleep. Some free plans stop quiet sites, and the next visitor waits for them to start. On Cloudzyro, sites never sleep, on the free plan too.
• Choose a host that doesn't take busy sites offline. On Cloudzyro Sites & Apps, a site over its monthly traffic is slowed, never switched off.
• Deploy safely. Use a host that checks a new version is healthy before sending visitors to it, and keeps old versions you can restore. On Cloudzyro Sites & Apps, a broken update never replaces a working version, and any earlier version can be restored in one click.
• Serve static files from a CDN. Cloudzyro sites sit behind Cloudflare, which also filters attacks.
• Keep your domain and certificate renewing automatically.
• Keep software updated, especially WordPress plugins, and test updates before they go live.
• Watch your host's status page. Cloudzyro's is at cloudzyro-status.statuspage.io, and the console shows a banner when something affecting you isn't working.

For the bigger picture on choosing a host, see what is web hosting and how to host a website.