When choosing hosting, most people compare price, disk space and plan names, but few read the service level agreement. Yet the SLA is the document that defines what happens when your site goes down: what the provider guarantees, how quickly support responds and what you receive if the guarantee is broken. Understanding it lets you act calmly during an outage instead of guessing.
What an uptime percentage really means
The headline number in an SLA is uptime, the share of time the service is available. The percentages look almost identical, but in minutes the gap is large. A 30-day month has 43,200 minutes, so the allowed downtime is easy to calculate.
| Uptime | Allowed downtime per 30 days | Per year |
|---|---|---|
| 99% | about 7 h 12 min | about 3.65 days |
| 99.5% | about 3 h 36 min | about 1.8 days |
| 99.9% | about 43 min | about 8 h 46 min |
| 99.99% | about 4 min | about 53 min |
The difference between 99% and 99.9% is tenfold. A business card site may not notice a few hours a month, but an online shop loses real orders. That is why how uptime is measured matters more than the number itself. Some providers only check that the server is powered on, others test the web server, database and mail separately. A five-minute check interval can also miss short one- or two-minute outages entirely.
What the guarantee usually excludes
Every SLA has an exclusions section, and most disputes start there. Typical exclusions include:
- planned maintenance announced in advance;
- errors caused by the customer's own code, plugins or settings;
- exceeding plan limits, such as a full disk or process limit;
- an expired domain or DNS records changed incorrectly by the customer;
- large-scale attacks and force majeure.
This is reasonable: a provider answers for what it controls. If an outdated WordPress plugin throws a 500 error, the server is fine and the fault is in the site code. What you should check is that the list is not so broad that it empties the guarantee, for example a vague clause like "any cause beyond the provider's control".
Response times and priorities
A good SLA also defines support response times. Distinguish first response time from resolution time: the first confirms your ticket is being handled, while resolution depends on complexity. Tickets are usually prioritised, so a site that is completely down is handled before a mail configuration question. Including the domain, the exact time, the error text and what you already checked speeds things up considerably.
How compensation is calculated
Many expect a full refund when the guarantee is missed. In practice hosting SLAs usually grant service credits on your balance, capped at a share of the monthly fee, for example a set percentage for every 0.1% below the target. The exact figures differ between providers, so read them in the agreement itself. Credits are often not automatic either: you must request them within a set period and state the date and duration of the outage.
This is where your own monitoring pays off. Imagine a small clinic in Samarkand with an online appointment form. Patients call to say the form fails, but nobody knows when it started or how long it lasted. A simple external monitor would have logged exact times and response codes. Even a basic cron check is better than nothing:
*/5 * * * * curl -s -o /dev/null -w "%{http_code} %{time_total}\n" https://example.uz/ >> ~/uptime.log
If this runs on the same server, it stops when the server does, so an external service is more reliable.
Questions to ask before you sign
Which services does uptime cover and how is it measured? How much notice is given for planned work? How and by when must credits be claimed? Are backups part of the SLA or a separate service? Asking these in writing is completely normal, and sayt.uz customers can raise them through a support ticket in the cabinet. An SLA does not prevent outages; it sets the rules for when they happen. Real reliability comes from the provider's guarantees, your own backups and the quality of your code working together.