· 2 min read

Uptime is the promise that your server responds. It's written as a percentage, but you feel it in minutes: 99.80% a month means at most 87 minutes when the site can be unavailable. Over a year, almost 17 hours. The difference between 99.80% and 99.99% looks small on paper, but it's 83 minutes a month — a whole morning of lost orders for an online store.

Where the percentage comes from

A site goes down for three reasons, in order of frequency: power, cooling and the Internet connection. That's why a data center should be judged by what's behind it, not by what the offer says.

  • Power — two medium-voltage lines and two transformers of our own, plus industrial UPS units and a generator. When one fails, the other takes over the load without the servers noticing.
  • Cooling — an N+1 redundant system: there's always one more unit than needed, so a single failure doesn't stop the room's air conditioning.
  • Internet — redundant connectivity from two different providers. If one has a routing problem, traffic goes through the other.

What it means for your site

Redundancy in the data center keeps the server running, but the rest depends on how the site is built: a heavy theme, a plugin that locks the database or a backup running at peak time can take a site down without the server being at fault.

That's why we monitor servers continuously and run daily backups with Veeam, with point-in-time recovery: if something breaks at site level, you don't wait to rebuild from scratch.

Questions to ask any provider

  1. Is uptime guaranteed in the contract, or just promised on the website?
  2. What happens if it isn't met — is there compensation, and how?
  3. How often are backups made, and how long does a real restore take?
  4. Does the provider own the data center, or rent it in turn?

With us the answers are the same every time, because the servers sit in our own data center, not someone else's.

← All articles

Read next