
Splitting a 30-seat office into three VLANs without a downtime window
The printers were the hard part, as usual. A migration order that let us move one department per evening and roll back in under a minute.
Since 2019 — independent
We are two people who keep the servers, networks and backups running for a handful of companies in southern Estonia. Everything we learn the hard way ends up written down here, in plain language, with the commands that actually worked.

Latest notes

The printers were the hard part, as usual. A migration order that let us move one department per evening and roll back in under a minute.

Restic to a local NAS, then to cold storage, with append-only keys so a compromised machine cannot delete its own history. Includes the restore drill we run twice a year.

Four checks per site, one escalation path, and a rule we stole from a client's night shift: if an alert cannot be acted on before morning, it is not an alert.
How we work
01
Most of our work starts with an undocumented network someone else built and left. First month is always mapping, labelling and writing down what exists.
02
Not for ideology — for the ability to read the config, script it, and move it to another box when the vendor changes its pricing.
03
Every credential lives in the client's own vault. If you stop working with us on a Friday, nothing breaks on Monday.
Said about us
They spent the first two weeks drawing our network on paper before touching anything. Nobody had ever done that here.
The restore test is the part I did not expect. We now know exactly how long it takes to get the accounting share back: fifty minutes.