Your host promises 99.9% uptime, and it sounds like "never down." It isn't. So what does 99.9 uptime mean in real time? Your site can be offline for about 43 minutes a month while the host still keeps its promise.
Quick Answer: 99.9% uptime allows 0.1% downtime. As of September 2026, that works out to 1 minute 26 seconds a day, 43 minutes 12 seconds in a 30-day month, and 8 hours 45 minutes 36 seconds a year. Stepping up to 99.99% cuts the monthly allowance to about 4 minutes 19 seconds. Most SLAs pay only a small service credit when a host misses its target.
Last reviewed: September 2026, with every figure recalculated for a 30-day month and a 365-day year.
Our team resolves 20 to 30 client site problems every day, and "is it down, and whose fault is it?" comes up constantly. This guide uses the same maths we walk customers through.
99.9% Uptime in Minutes: Per Day, Month and Year
Three nines leaves 0.1% of every period for downtime, and each extra nine divides that allowance by ten.
| Uptime level | Allowed downtime per day | Allowed downtime per 30-day month | Allowed downtime per year | |---|---|---|---| | 99% (two nines) | 14m 24s | 7h 12m | 3d 15h 36m | | 99.5% | 7m 12s | 3h 36m | 1d 19h 48m | | 99.9% (three nines) | 1m 26s | 43m 12s | 8h 45m 36s | | 99.95% | 43s | 21m 36s | 4h 22m 48s | | 99.99% (four nines) | 8.6s | 4m 19s | 52m 34s | | 99.999% (five nines) | 0.9s | 26s | 5m 15s |
The monthly figure shifts with the calendar. A 31-day month allows 44 minutes 38 seconds, February allows 40 minutes 19 seconds, and calculators built on an average month show about 43 minutes 50 seconds.
The 99.9 vs 99.99 uptime difference is bigger than the decimal suggests. As of September 2026, a 99.9% promise allows 43 minutes 12 seconds of downtime in a 30-day month, while 99.99% allows 4 minutes 19 seconds, ten times less. Over a year, that's 8 hours 45 minutes against 52 minutes. Google's SRE team explains why each extra nine costs real engineering effort in its chapter on embracing risk.
Would I even notice 43 minutes of downtime?
Maybe not, and that's the trap. Downtime rarely arrives as one block; it shows up as six 7-minute blips overnight, which a visitor abroad hits mid-afternoon. Your allowed downtime per month gets spent in pieces, and you usually hear about it only when a customer emails a screenshot.
Uptime SLA Explained for Hosting: The Four-Line Check
An uptime percentage is only as strong as the four lines of small print behind it. Hostaccent's engineers use a quick test we call the Four-Line SLA Check whenever a customer asks us to compare guarantees, and it takes about 5 minutes on any host's terms.
- The measurement window. Is uptime measured monthly or yearly? A yearly 99.9% SLA lets a host spend the full 8 hours 45 minutes in one bad afternoon and still pass. Measured monthly, the same outage breaches after 43 minutes.
- The definition of "down". A ping to the server, or a check that your page actually loads? A server can answer pings while every site on it throws errors.
- The exclusions. Look for scheduled maintenance, DDoS attacks, faults caused by your own code or plugins, and upstream network or DNS problems.
- The claim rules. Check who measures the outage, how many days you have to claim, and the cap on what you can receive.
Scheduled maintenance is the exclusion that moves the numbers most. If an SLA excludes a 2-hour monthly maintenance window on top of 43 minutes of unplanned downtime, a 99.9% host can keep your site offline for about 2 hours 43 minutes in a 30-day month and still report full compliance.
Pro Tip: Don't rely on the host's status page. Run your own external HTTP check at 1-minute intervals from at least two regions, and log incidents in UTC. A 5-minute check interval can miss a 4-minute outage entirely, and without timestamps you have nothing to support a credit claim.
Still working through this server issue?
Send the symptoms, error output, and what you have already tried. We can work with Hostaccent services or infrastructure hosted with another provider.
What Is an Uptime Guarantee Credit Actually Worth?
Usually a few dollars. Credits are almost always a slice of that month's hosting fee, commonly tiered from around 5-10% for a small miss up to the whole month for a severe one, and they arrive as account credit rather than cash.
Run the numbers on a VPS costing $7.99/mo. A 10% credit for a missed month is $0.80. Now picture a small shop earning $1,500 a day, roughly $1.04 a minute. Forty-three minutes offline costs it about $45 in orders, before counting ad spend that kept running. The credit covers under 2% of that loss, which makes it a signal of the host's confidence, never insurance.
Search visibility takes a hit only when errors repeat. Google's documentation on how HTTP and network errors affect crawling explains that 5xx responses slow Googlebot down, and URLs that keep failing can eventually drop out of the index. One 20-minute outage won't sink rankings, but a week of evening crashes can quietly explain why a website stops getting traffic.
When you take a site down on purpose, return a 503 Service Unavailable status with a Retry-After header instead of a normal page saying "back soon". Crawlers then treat the gap as temporary.
What Does 99.9 Uptime Mean for Your Actual Website?
It describes the host's server and network, not your site. The guarantee stops at the machine; your visitors experience everything stacked on top.
According to Hostaccent's support queue (September 2026), our engineers resolve 20 to 30 client issues a day, and a large share of "my site is down" reports involve a healthy server running a failing site: a full disk, exhausted PHP-FPM workers, an expired SSL certificate, or a DNS record edited in the wrong place. We've handled disk-full and inode-full emergencies where every site on a server went offline together while the machine answered pings perfectly. A ping-based SLA would have scored that 100%.
Dependencies also multiply. A site that needs a 99.9% server, a 99.9% database service and a 99.99% DNS provider all working at once has a combined ceiling of about 99.79%, which allows roughly 1 hour 31 minutes of downtime in a 30-day month.
The mistakes we see most often:
- Counting slow as up. A page that takes 20 seconds is effectively down for most visitors, so check your website speed alongside availability.
- Blaming the host for every browser error. A "This site can't be reached" message is often DNS or the visitor's own connection.
- Skipping the logs. Your cPanel error log usually shows whether PHP, the database or the web server failed first.
- Panicking about email. Sending servers retry for days (Postfix defaults to 5), so short outages rarely lose mail. A 550 email bounce is a different, permanent rejection.
Insider Insight: Across the 4,000+ site migrations Hostaccent has handled, the step people most often skip is lowering DNS TTL. Drop it to 300 seconds a full day before the move, and the cutover gap shrinks from hours of split traffic to a few minutes.
Key takeaways:
- 99.9% allows 43m 12s per 30-day month; 99.99% allows 4m 19s.
- Check the window, definition, exclusions and claim rules before trusting any percentage.
- Credits are small, so your own 1-minute monitoring is the real protection.
Your Next Step: Judge Hosts by Minutes, Not Decimals
If you arrived asking "what does 99.9 uptime mean," you now know: 43 minutes a month, backed by a credit worth less than a coffee. You can police that yourself, or start where the stricter promise is already written down. Hostaccent (hosting since 2016, UK-incorporated 2018) backs its VPS plans with a 99.99% uptime guarantee, free 30 Gbps DDoS protection and 24/7 help from our own engineers. One honest caveat: you get full root access but no control panel licence, so if you'd rather not manage a server, start with shared hosting. Otherwise, start on the Basic VPS plan at Basic, $7.99/mo, renewing at $7.99/mo, with a 30-day money-back guarantee.
Frequently Asked Questions
What does 99.9 uptime mean in hours per year?
It means up to 8 hours 45 minutes 36 seconds of downtime in a 365-day year, or about 8 hours 46 minutes if a calculator uses 365.25 days. That sounds small, but it's a full working day. Whether it hurts depends on timing: an hour at 4 a.m. local time barely registers, while the same hour during a product launch or sale can cost more than a year of hosting fees.
Is 99.9% uptime good enough for a small business website?
For a brochure site or blog, usually yes. Forty-three minutes a month, spread across short blips, rarely costs a small business real money. It becomes a problem when the site takes orders, bookings or paid ad traffic, because every minute offline is lost revenue and wasted ad spend. In that case, look for 99.99% backed by a monthly SLA, and run your own 1-minute monitoring so you know the real figure.
How much more reliable is 99.99% than 99.9% uptime?
Ten times more reliable on paper. A 99.9% guarantee allows 43 minutes 12 seconds of downtime in a 30-day month; 99.99% allows 4 minutes 19 seconds. Across a year the gap is 8 hours 45 minutes versus about 52 minutes. Reaching four nines normally requires redundant hardware, fast automated alerting and careful maintenance, so check that the host's SLA is measured monthly and pays credits before you treat the extra nine as meaningful.
Does scheduled maintenance count as downtime in an SLA?
Usually not. Most hosting SLAs exclude announced maintenance windows, which means a site can be offline for longer than the percentage suggests while the host still reports full compliance. Read how much notice the host must give, how long windows can last and how often they happen. A provider that does rolling or live maintenance, with little or no visible downtime, is worth more than one with a bigger number and generous exclusions.
How do I claim an uptime SLA credit from my host?
Open a support ticket within the claim window, which is often 7 to 30 days after the month ends. Include the date, start and end times in UTC, the affected URL and evidence from an external monitor, such as screenshots or exported logs. Credits normally arrive as account credit against your next invoice rather than a cash refund, and they're capped, frequently at one month's fee for the affected service.
Discussion
Have a question or tip about this topic? Share it below — your comment will appear after review.