Things we learned the expensive way, written down

No vendor marketing and no reposted press releases. These are notes from our own engineers about problems we keep encountering in real environments, and what actually works here.

FeaturedSecurity·7 min read·4 August 2026

Backups are not a recovery plan: a restore-testing checklist

Almost every organisation we assess has backups running. Far fewer can demonstrate a restore. In three of the ransomware incidents we responded to last year, the backup job had been green for months and the data was unrecoverable, because nobody had ever tried. Here is the quarterly drill we run with clients, and the six failure modes it catches.

Read the full article

All articles

Latest from the engineering team

Networking·9 min

Choosing between MPLS, SD-WAN and plain internet in Iraq

Carrier realities here are not the ones in the vendor deck. A practical comparison based on sites we actually operate, including what each option really costs over three years.

Read

Cloud·6 min

What a Microsoft 365 tenant should look like on day one

Conditional access, MFA, retention, external sharing defaults and the ten settings we change on every tenant we inherit, with the reasoning for each.

Read

Infrastructure·8 min

Why your UPS keeps failing, and it is probably not the UPS

Earthing, harmonic distortion, generator transfer behaviour and battery temperature. Four causes we find repeatedly behind "faulty" UPS units that test fine on the bench.

Read

Security·11 min

Network segmentation for people who inherited a flat network

You cannot re-architect everything at once. A staged segmentation approach that reduces blast radius meaningfully without a big-bang cutover.

Read

Wireless·7 min

Wi-Fi design for density, not coverage

Why a lecture hall with full signal bars still fails at 300 users, and how to survey and size for concurrency instead of a green heat map.

Read

Procurement·10 min

Reading an IT quotation: twelve things to check before you sign

Warranty registration, licence terms, who owns the configuration, what "installation" includes, and the clauses that decide what happens when something breaks in year two.

Read

Operations·6 min

Monitoring alert fatigue: cutting noise by 70% without going blind

Dependency suppression, threshold tuning and maintenance windows. How we reduced one client's monthly alert volume from 14,000 to 4,100 without missing an incident.

Read

Applications·9 min

ERP projects fail in the workshop, not the code

The process-mapping phase everyone compresses to save budget, and why it is the single strongest predictor of whether the system gets used after go-live.

Read

Field·8 min

Building IT for remote sites: a checklist from the oil fields

Power, dust, heat, satellite latency and the four-hour drive to the nearest engineer. What we changed in our standard build after the first three deployments.

Read

Article pages are placeholders. The titles and summaries above outline the planned content.

One email a month. No product pitches.

A short digest of what our engineers ran into that month, plus any security advisories relevant to the region.