Hosting architecture
AWS, DigitalOcean, Hetzner or UK-based managed hosting, chosen on workload and data residency rather than habit. Sized for what you run, with headroom for a good month.
04 — Hosting, pipelines and resilience
Cloud and Infrastructure
Hosting, pipelines and recovery, sized to what you actually run.
The short version
Hosting that fits the workload, deployment pipelines that remove the drama, and backups you have actually tested.
Infrastructure is invisible when it is right and catastrophic when it is wrong. We size it to the workload rather than to a brochure, automate the deployment so that shipping is boring, and rehearse recovery before you need it. And we keep an eye on the bill, because cloud spend has a way of drifting upwards when nobody is looking.
What this covers
Six things we are asked for most.
Deployment pipelines
Git-based CI/CD with staging environments, automated tests, database migrations and one-command rollback. Releasing should be a non-event.
Backup and recovery
Off-site, encrypted, versioned and restored on a schedule to prove they work. A backup nobody has restored is a hope, not a plan.
Resilience and failover
Health checks, graceful degradation, queue-backed jobs and a documented recovery procedure with real numbers for how long it takes.
Cost review
A look at what you are paying for and what you are using. Right-sizing, reserved capacity and turning off the things nobody has touched in a year.
Environment parity
Local, staging and production built from the same definition, so "it worked on my machine" stops being part of the conversation.
What you end up with
Deliverables, not deliverable-shaped promises.
- Deployments that take minutes, not evenings
- A recovery plan with a tested time to restore
- Infrastructure defined in code and version controlled
- A hosting bill that matches what you actually use
Typical stack
- AWS
- DigitalOcean
- Hetzner
- Docker
- GitHub Actions
- Terraform
- Nginx
- Cloudflare
- Let's Encrypt
- Restic
How it starts
A thirty-minute call, then a short paid discovery that produces a written specification and a fixed price. If you walk away after discovery, the specification is yours to take to anyone.
The path
Same five steps, every time.
01
Conversation
Thirty minutes, no charge, no deck. You describe the problem in your own words and we ask the awkward questions. By the end of it we both know whether this is worth taking further.
You get: Outline of the problem, rough budget range, honest yes or no
02
Discovery
A short paid engagement where we sit with the people who will use the thing. We map the current process, find the edge cases that break naive designs, and write it down.
You get: Written specification, wireframes, milestone plan, fixed price for phase one
03
Build
Two-week cycles with something you can click at the end of each one. You see progress continuously, which means course corrections cost days rather than months.
You get: Working software on a staging URL, updated every fortnight
04
Launch
Data migrated and reconciled, team trained, rollback rehearsed, go-live in a planned window. We stay close for the first month while real use finds the last few edges.
You get: Live system, trained team, documentation, thirty days of included support
05
Operate
Monitoring, patching, backups and a monthly written report. Or a clean handover to your own team with everything they need. Both are a good outcome for us.
You get: Retained support, or full handover with documentation
Questions
The ones people actually ask.
Not on the list? Email us — we would rather answer it now than after you have signed something.
Yes. UK and EU regions are available across the providers we work with, and we will put the region and the sub-processor list in writing so your compliance file is straightforward.
Not necessarily. Plenty of sites run happily on shared hosting for years. It becomes a problem when you need staging, background jobs, real backups or predictable performance. We will tell you which side of that line you are on.
No. Accounts are in your name, billed to you, with us added as a user. If you leave, you keep the infrastructure and we remove our access.
A straightforward site is usually a rehearsed evening. Anything with live transactions, large media or DNS complexity is planned over a couple of weeks with a short, agreed cutover window.
Next
Software Development
Custom applications for the parts of your business that off-the-shelf software will not fit. Built properly, documented, and handed over in full.
Start here
Tell us what you are trying to build.
A short conversation is usually enough to tell whether we are the right team for the job. No sales script, no obligation.
We reply within one working day. No sales sequence, no pressure.