PixelCodeLab

The platform

Built once. Watched every five minutes.

Every site the studio builds runs on the same setup: one server, one shared library of code, and a second machine somewhere else whose only job is to notice when something breaks. This page says exactly what that means, because most studios will not.

Uptime, 30 days
97.31%
across every site we host
Sites watched
16
15 answering right now
Checks, 30 days
106,238
one every five minutes, per site
You hear about it in
10 min
two failed checks in a row
Measured by the monitoring machine, not by the sites themselves, and refreshed every ten minutes.
Read the live status page ↗ What it costs

The server

One machine in the United Kingdom, rented and run by the studio, not resold to you.

  • Each site runs in its own container, so one busy site never slows another.
  • Cloudflare sits in front of every site and absorbs the traffic a small server cannot.
  • A deploy runs the test suite first. A red suite never reaches the server.
  • The database is dumped before every migration, so a bad change has a way back.
  • A new release is built beside the live one and switched in only once it answers. If it fails, the site switches back in seconds.

Watching

A second machine, at a different company, checks every site from outside the network it lives on.

  • Every five minutes: the page, how long it took, and the certificate behind it.
  • An alert goes out on the second failure in a row, which is about ten minutes.
  • The alert reaches Luke by email, through a mail provider that shares nothing with the server.
  • If your site is down for 15 minutes, you get an email saying so. Once it is back, Luke sends you a short report of what happened and what changed.
  • A public status page carries 90 days of history, and it stays up when the server does not.
  • A dead machine says nothing, so a timer elsewhere raises the alarm when the watcher goes quiet.

Backups

A copy of your site leaves the server every night and lands somewhere else.

  • The database is dumped nightly and kept for 14 days.
  • A copy is sent off the server to the monitoring machine.
  • The monitor raises a fault when no backup arrives for 36 hours, because a backup that stops quietly is how people find out too late.
  • Files you upload live on the same disk as the site and travel with it.

Security

The boring parts, done the same way on every site.

  • HTTPS everywhere, with certificates that renew themselves and a warning 14 days out when a renewal fails.
  • Your admin panel takes a passkey or a code as a second factor.
  • Passwords and keys live in a vault the studio hosts, never in an email or a chat message.
  • The settings a site needs to run are encrypted in the code repository.
  • Nobody reaches the server over the open internet. Administration runs on a private network.
  • Framework and library updates land through the shared library, so a fix reaches every site at once.

Password vault

Every site gets its own vault for passwords and keys, at vault. and your domain.

  • Only you hold the key that opens it. The server stores the vault encrypted, and the studio cannot read an entry.
  • The studio invites your team and removes people who leave, so access ends the day they do.
  • It works with the Bitwarden apps on phone, browser and desktop. The server is Vaultwarden, which is open source, and support comes from the studio.
  • Each vault runs in its own container and is backed up every night with everything else.

Email

Your address stays yours, and the messages your site sends arrive.

  • Mailboxes run on the estate mail server, with webmail and your usual phone or desktop app.
  • Incoming mail is checked for spam, for known viruses and for phishing links before it reaches you.
  • Every domain is signed with DKIM and SPF, which is what keeps a form message out of the spam folder.
  • Enquiries from the site are stored in your panel as well as emailed, so a lost email is not a lost customer.
  • Alerts about an outage leave through a different provider on purpose. A mail server on the broken machine cannot report that the machine is broken.

And in your admin panel

  • Traffic dashboard. Visitors and pages, without Google Analytics or a cookie banner.
  • Enquiries inbox. Every form message stored and emailed, spam throttled.
  • Reviews on the site. Google or Trustpilot reviews synced nightly and marked up for search.
  • Search health. Titles, descriptions, sitemap and structured data audited on every deploy.
  • Checked every five minutes. Uptime and certificate monitoring. If your site is down for 15 minutes, you hear from Luke.
  • Nightly backups. Database dumped before every deploy and every night, kept for 14 days.
  • Your own admin panel. Edit words, images, prices and pages yourself.
  • Care plan. Updates, fixes and small changes each month, from GBP 49.

Every site is checked every five minutes from outside the server, by a machine at another company. How a site is run · Live status page ↗

Your site can run on this too.

Care plans start at £49 a month and cover the server, the watching, the backups and the updates. Moving an existing site over is a fixed price.

Talk to Luke See pricing