Real case

WordPress Migration from OVH to AWS

Revista GODOT front page on its new server.

The Revista GODOT website used to go down every day. Now it doesn’t go down, and it costs 30% less.

Context

Client
Revista GODOT
Industry
Media and promotional gifts
Size
Grupo Editorial

The challenge

The Revista GODOT website was going down several times a day, sometimes for more than an hour. We looked into it and found two causes. When lots of people visited at once, the server ran out of memory. The site has an enormous number of images. And some of WordPress’s automated tasks requested a lot of data all at once, so much so that they brought the site down even with few visitors.

The outages had to be reduced, and all the information had to be easy to recover if it was lost. On top of that, the site had accumulated a huge amount of data, and nobody was maintaining it on a regular basis: any change could trigger a new outage.

What we built

We moved the Revista GODOT website to a new server with more capacity, registered in the client’s own name: the server belongs to them, not us. We then changed when WordPress runs its automated tasks: no longer on every visit, but once an hour. As a result, the site went from several outages a day to one a week. And whenever it did go down, we had it back up in under fifteen minutes.

Later on, we put its images on copies distributed around the world, so its server no longer has to send them one by one. Since then, zero outages.

We migrated the web store from its previous provider to the new server in one week.

Results

  • 30 % in server savings, with better performance than before
  • 0 outages, previously several a day
  • 1 week per website migrated

For technical readers: how it is built
1 2 3 4
How the system is built inside.
  1. The site leaves the previous server.
  2. It enters a new server, in the client’s own account.
  3. A scheduled task sets the time for automatic processes.
  4. They run once an hour, not with every visit.

The architecture of a migration like this is deliberately simple:

  • Lightsail hosts the site on an AWS instance with a fixed size and a fixed monthly price.
  • Apache, PHP and MySQL make up the classic stack on which WordPress runs on that instance.
  • Amazon S3 stores files outside the server.

A small decision with a big impact. WordPress cron no longer runs on every visit: a system cron job triggers it once an hour.

What it was built with and how long it took

  • Apache
  • AWS S3
  • Lightsail
  • MySQL
  • PHP
  • WordPress

Duration: 1 week per migration

Related services