Infrastructure and cloud

What we do

We work with Linux servers: physical machines on your premises, rented hardware in a data centre, or virtual machines in Google Cloud. The scope depends on what you already have:

We pay particular attention to ransomware, because it goes after the backups too. At least one copy cannot be deleted from the main server, either through separate credentials or storage where written data cannot be altered for a set period. Even after a breach, there is something to restore from.

• Installing and configuring servers for web applications, databases, warehouse and production systems.
• Moving existing systems from old hardware to new, from an on-site server to the cloud, or back again.
• Backups on the three-two-one rule — three copies, two kinds of media, one off-site — with regular test restores.
• Monitoring of load, disk space, certificates, services and application response times, with alerts.
• Scheduled security updates, firewalls, key-only access and logged sign-ins.
• Google Cloud and BigQuery for data that has to be analysed at volume or shared between systems.

Cloud or your own server

The cloud is not automatically better. For a warehouse that must keep running when the internet goes down, a local server with a cloud copy is often the more sensible choice. For an online store whose traffic doubles over the holidays, or for data that several sites need to reach, the cloud has clear advantages: flexible capacity, managed databases, and hosting in an EU region.

We work out the numbers with you: what each option costs per month, what happens in a failure, who maintains what. In Google Cloud we set up budgets and cost alerts from day one so the monthly bill is never a surprise. The answer is often a mix — the production system stays on site while backups, reporting and analysis move to the cloud.

We also try not to tie you to a single provider. Applications are packaged in Docker containers and the configuration is written down, so the system can be brought up on another server or another cloud without being rewritten. Done from the start, that insurance costs almost nothing.

Validation: knowing it works, not assuming

A backup that has never been restored is not a backup; it is a hope. So at agreed intervals we restore the system from backup onto a separate machine and check that it starts and the data is intact. The result is recorded: when the test ran, how long the restore took and what was checked.

The same goes for launching a new system or moving an old one. Before the switchover we draw up a checklist — users can sign in, reports come out, integrations pass data, certificates are valid — and go through it with someone from your side. If any item fails, the switchover is postponed and the old system keeps running. We would rather be a day late than a day down.

BigQuery and data for analysis

When heavy reports start slowing down your production database, or when you need data from the warehouse, the tills and the online store in one place, we move it into BigQuery. A report across years of history takes seconds there, without loading the system your staff are working in. The same data can then feed AI agents and management dashboards.

We set up the scheduled transfers, the access rights and cost controls. BigQuery charges by the volume of data processed, and a badly written query can cost more than it should.

What you get, and what comes next

A written description of your infrastructure — which servers exist, what runs on them, where the backups are and who has access — plus a recovery procedure written so that someone else could follow it. Passwords and keys stay with you. Support covers monitoring, updates, test restores and incident response, 24/7 for critical systems. Once a year we sit down together and check whether the setup still fits what you need.