You launch a promotion on a Friday afternoon and expect a steady stream of visitors for the next 48 hours. By Monday morning the checkout page is offline and every alert on your phone is flashing red. The host's marketing page still proclaims "99.9 % uptime guaranteed." How long could that outage have legally been allowed under the claim?
The Numbers in Real Hours
Uptime percentages are simple fractions of a year. A non-leap year has 8 760 hours. Multiply that by the shortfall from 100 %:
| Uptime | Allowed downtime per year | |-------|----------------------------| | 99 % | 87 hours 36 minutes | | 99.9 % | 8 hours 46 minutes | | 99.99 %| 52 minutes 33 seconds | | 99.999 %| 5 minutes 15 seconds |
The most common promise, 99.9 %, gives you almost nine hours of downtime every 365 days. Tightening the figure to 99.99 % shrinks that window to just under an hour. Those are hard limits - they do not account for any exclusions that hosts often embed in their service level agreements (SLAs).
What Real-World SLAs Usually Exclude
When a host publishes an SLA, the word "uptime" typically means "the server process is running and reachable." Most contracts carve out several categories that never count against the promised percentage:
- Planned maintenance - scheduled updates, kernel patches, or hardware swaps are treated as a normal part of operations. The host announces the window, but the minutes are not deducted from the SLA.
- DDoS mitigation - large traffic floods are classified as external attacks. The host may throttle or filter traffic, and those minutes are excluded.
- DNS failures - if the authoritative DNS servers do not respond, the underlying web server may still be healthy. The outage is recorded as a DNS issue, not a web-server outage.
- Third-party services - payment gateways, email providers, or CDN nodes are outside the host's control and are rarely counted.
Because of these exclusions, a host can technically meet a 99.9 % figure while you experience several hours of real-world downtime due to a DNS glitch or a scheduled kernel reboot that lasts longer than announced.
What TrueCore Actually Promises
We do not advertise a headline uptime percentage because the figure hides more than it reveals. Instead we focus on transparency and concrete actions:
- Live status page -
https://truecorehosting.com/status.htmlis refreshed every 60 seconds by our internal monitorflame-watchman. The page shows the health of each node (ember, spark, litespeed) and the status of DNS, email, and database services. - Advance notice of maintenance - All planned work is announced at least 48 hours in advance. Typical kernel updates and reboots finish in 2-5 minutes, keeping the window well below the 8-hour annual allowance for 99.9 % uptime.
- Short, scripted maintenance - Our deployment scripts run under
flame-bubblecontainers, ensuring that each change is isolated and can be rolled back instantly if something goes wrong. - Redundant DNS - Three authoritative nameservers run on separate nodes in Germany, the US, and Bulgaria. A single DNS failure cannot take your site offline.
- Compensation policy - Where we offer a service credit for an unplanned outage, it is a goodwill gesture rather than a contractual formula, and we are honest that any such credit is modest relative to the real revenue an outage can cost you. We would rather keep windows short and communicate clearly than advertise a payout that pretends to make you whole.
We also encourage you to run your own monitoring. A simple curl against the status page can be scripted to alert you the moment a node flips to "down". For example:
#!/bin/bash
if curl -sSf https://truecorehosting.com/status.html | grep -q "node: down"; then
echo "TrueCore node down!" | mail -s "Urgent: Node outage" you@example.com
fiRunning this script on a cheap VPS or a local machine gives you an independent view that complements our public status.
How the Numbers Translate to Your Business
Take the earlier promotion scenario. If your site goes down for three hours on a 99.9 % host, you have used about a third of the yearly allowance in a single weekend. For a boutique that processes £200 per hour, that outage costs £600 in lost sales. The host's typical SLA credit would be around £2, which does little to cover the real impact.
If you were on a plan that promises 99.99 % uptime, the same three-hour outage would break the contractual guarantee. In that case many hosts would be forced to provide a higher credit or negotiate compensation. The tighter guarantee is more than a marketing gimmick - it forces the provider to keep maintenance windows short and to invest in redundancy.
What to Watch in Your Own Monitoring
Independent checks give you three data points that matter more than a headline percentage:
- Response code - 200 OK versus 500 Internal Server Error tells you if the web stack is alive.
- Time to first byte (TTFB) - Anything above 5 seconds is effectively a bad experience even if the server returns 200.
- DNS resolution time - A query that takes longer than 100 ms may indicate a problem with the authoritative nameserver.
Tools like UptimeRobot (5-minute checks), Freshping (1-minute checks), and Better Uptime (free tier includes response time) can be set up in minutes. Pair them with a webhook that posts to a Discord channel or sends an SMS, and you have a real-time alarm system that does not rely on the host's own statistics.
Bottom Line
"99.9 % uptime" translates to 8 hours 46 minutes of permitted downtime per year. That figure alone does not guarantee that your site will stay reachable when you need it most. Look for hosts that publish live status, give ample notice of maintenance, keep windows short, and offer a clear credit policy. TrueCore follows that approach, and we back it with concrete tools like flame-watchman and a three-node DNS fleet. Use independent monitoring to verify the service you pay for, and you will have the data you need to make an informed decision.