This is information gathered from Wikipedia to hopefully explain to you what it really means to have reliability and uptime...

The availability of a website is measured by the percentage of a year in which the website is publically accessible and reachable via the internet. This is different than measuring the uptime of a system, as uptime refers to the system itself being online, however does not take into account being able to reach it – as in the event of a network outage.

The formula to determine a system’s availability is relatively easy: Total time = 365 days per year * 24 hours per day * 60 minutes per hour = 525,600 minutes per year. To calculate how many minutes of downtime your system may experience per year, you can take your uptime guarantee and multiply it by total time in a year.

In this example I'll use 99.99%: (1 - .9999) * 525,600 = allowable minutes down per year.

The following table shows the translation from a given availability percentage to the corresponding amount of time a system would be unavailable per year, month, or week.

A hosting provider’s SLAs may include a certain amount of scheduled downtime per year so that they can perform maintenance on the systems. This scheduled downtime is often excluded from the SLA timeframe, and needs to be subtracted from the Total Time when availability is calculated. Depending on the verbiage of an SLA, if the availability of a system drops below that in the signed SLA, a hosting provider often will provide a partial refund for time lost.

The percent uptime advertised by a web host is often a bad metric for determining a hosts quality. With this metric, a scheduled downtime from 2AM-3AM will be counted as just as bad as an unplanned downtime from 5PM-6PM.