Articles & guides
We write down what we learn on projects — practical guides, no slogans, no hidden pitch.
All articles
Articles shown: 12
When the international link goes down, what does your infrastructure do?
Most systems don't fail on the day of the outage; they fail at the first update, deployment or restart after it. The hidden dependencies, and a local replacement for each.
- Security3 min read
Why centralised logging is the first real security step
Before buying any expensive security tool, answer one simple question: if a breach happened right now, how long would it take you to notice?
- DevOps3 min read
Multi-stage Dockerfiles: from 1.2 GB down to 90 MB
A bloated image doesn't just take up space — it inflates deploy time, bandwidth cost and attack surface. The fix is simpler than you'd expect.
- Security4 min read
Two-step sign-in for internal systems: SMS, an app, or a hardware key?
Sooner or later a password leaks; the question is what stops an attacker after that. Three common methods, the weak point of each, and the order to roll them out.
- Security3 min read
Backups that ransomware can't reach
Modern ransomware goes for the backups first and the data second. The 3-2-1-1-0 rule, and the mistakes that quietly defeat it.
- Automation4 min read
AI on your own server: when does a local model make sense?
A local model means data never leaves the organisation and an outage doesn't stop the work. But it isn't good at everything: where it fits, and what hardware it needs.
- Automation3 min read
Human approval in automation: where the machine decides and where a person does
Full automation isn't always the right goal. A simple test for which steps need approval, and how to design one that doesn't become a reflex click.
- Guide3 min read
The first hour after you find a breach: what to do, and what not to
In the first hour, instinct does the most damage: switching servers off, deleting files, announcing it over an email system the attacker may be reading. A checklist for that hour.
- Automation3 min read
Build your first AI agent with n8n
A good agent is not magic: it's a simple loop of tools, decisions and oversight. Here we build one from scratch.
- Hardware2 min read
Why a dedicated appliance instead of a cloud service?
The cloud is excellent for a great many things. For security monitoring, three specific reasons put local hardware ahead.
- Guide3 min read
Hardening a Linux server in ten practical steps
The checklist we run on every project. No extra theory — just the things that genuinely reduce attack surface.
- DevOps3 min read
What an SLO is, and how to set your first one
"99.9% uptime" with no precise definition is a slogan. An SLO is that definition — and it's useful for far more than a monthly report.