Not resold space on someone else's platform. We own the infrastructure, so when something needs changing we change it rather than raise a ticket.
← Back to servicesSERVICE / 03
Infrastructure you stop thinking about.
Hosting is the part nobody notices until it fails, and then it is the only thing anybody notices. We set up foundations that are boring on purpose, and we measure them instead of assuming.
01 / The brief
Most performance problems are configuration, not code.
Compression that never reaches your JavaScript. Cache headers that make every visit a first visit. A revalidation that returns 200 instead of 304. None of it shows up in a code review, and all of it is measurable in an afternoon.
02 / What's included
What's included
A panel built here that gives you access to your own server — status, resources, logs and the controls you actually need, without a support queue in between.
Both are in-house services that are set up and tested before you go live, not add-ons bolted on after the first incident.
Registration, transfer and renewal handled with the rest of it, priced well below what the big registrars charge once the first year ends.
Functions and managed services where bursty demand suits them, real servers where predictable load makes them cheaper. We are not attached to either.
Alerts tied to what visitors experience, and a named person who answers when one fires.
03 / How it goes
How it goes
- 01Measure the current state
Real transfer sizes, real cache hit rates, real time to first byte — before touching anything.
- 02Fix the cheap things first
Compression, headers and image weight usually return more than a bigger server, at a fraction of the cost.
- 03Automate the release
One command, same every time, with the database migration and the cache invalidation in the right order.
- 04Prove it again
The same measurements after the change, so the improvement is a number rather than a feeling.
04 / How we build
How we build
Before-and-after numbers on every change. A cache is only working if the second visit is cheaper, and that is testable.
Every service added is a service to patch, monitor and reason about at three in the morning.
Backups you have restored from at least once. Untested backups are a belief, not a plan.
05 / Seen in practice
Lizaria
Lizaria runs Shopware, MySQL, Redis and a Nuxt front end in separated containers behind nginx, with Redis split by role and a cache dashboard that distinguishes "available" from "actually effective".