Every hosting company says it’s reliable. Almost none will show you what actually happens the moment a server dies.
So we did the uncomfortable thing. On a live WordPress site — serving real traffic and taking WooCommerce orders — we deliberately broke our own platform, and we recorded every single request a visitor made while we did it.
Here’s exactly what happened.
Test 1: We killed a running web server, mid-request
Not a graceful shutdown. We pulled the server out while it was in the middle of serving requests — the worst case — with 60 people browsing at once.
The site never went down. It was already running on more than one server, so the load balancer simply stopped sending traffic to the dead one. Out of nearly 5,000 requests during the test, exactly ten hung briefly at the moment of the kill. Every other request loaded normally. No error pages. No white screens.
A visitor’s worst-case experience: one page that took a few seconds longer than usual, once.
Test 2: We took a whole machine offline
Killing one server is one thing. Losing an entire physical machine is another — that’s a real datacenter event.
So we drained a whole machine out from under the site while it was serving. The site moved to a healthy machine on its own. A probe checking the homepage every single second recorded zero failures from start to finish. The move took about the time it takes to read this paragraph.
Test 3: We shut down the database in the middle of a checkout
This is the one that keeps hosting engineers up at night — and it’s the one that matters most if you run a store.
When your database fails over, the danger isn’t a “down” page. It’s a customer who clicked Place Order, got charged, and whose order vanished into a half-written state. A lost sale that neither of you knows about until they email asking where their product is.
So we set up a probe writing a continuous stream of orders through the database, and then we shut the primary database down.
A standby was promoted automatically. Writes paused for a few seconds — and then every one of them resumed and committed. Afterward, we checked the data row by row: not a single confirmed write was lost. No stranded orders. No duplicates.
That’s the number we care about most. Not “how fast” — but “did anything get lost.” Nothing did.
We also made sure we can get your content back
Speed of recovery is only half the story. The other half is whether your actual content is safe — and recoverable.
So we took a real site’s nightly backup and restored it into an isolated environment, then checked it against the live site. Every post, every user, every setting — all 409 posts, down to the titles — came back an exact match. Not “the backup exists.” The backup works, and we’ve watched it work.
Being straight with you
No host can promise literally zero downtime or zero data loss. Anyone who tells you they can is selling you something.
What we can tell you is this: we don’t guess about reliability. We test it — by deliberately breaking our own platform, on a live site, and measuring what a real visitor would actually experience. In these tests, failures were measured in seconds, and no confirmed data was lost.
That’s the standard we hold ourselves to. Not a marketing number on a slide — something we’ve watched happen, on purpose, more than once.
If that’s the kind of platform you want your WordPress site on, we’d love to have you in our beta.