Disaster recovery

What it means

Disaster recovery is a prepared route back to operation after an outage - a failed server, ransomware, a fire, a lost login. It is not only IT: who decides, who gets called and how the business runs meanwhile all belong to it.

The plan rests on two numbers. RTO says how quickly the service must be back; RPO says how much data may be lost. Those two values are what the whole arrangement costs money for.

Disaster recovery is not the same as data backup. A backup is only one ingredient of a restore; the plan also covers the order of systems, replacement hardware, access credentials and how work carries on while the restore runs. A company can have flawless backups and still stand idle for a week.

How it works in practice

For the accounting function what matters is that the data exists in two independent places and can be restored without one particular person. A backup in the cloud handles that technically; the rest is documentation and access.

In the vendor contract it pays to have restoration inside the SLA alongside availability (uptime). Without an agreed recovery time, "we will restore as soon as we can" is unenforceable.

A small company needs no more than a one-page plan: the systems in order of importance, where the backups are, who holds the access and whom to call. When ransomware hits a client, the deadlines that will not wait come first - paying wages and filing the VAT return - and only then the restore of individual computers.

What to watch out for

A plan never rehearsed generally does not work. A test restore once a year exposes missing passwords, incomplete backups and dependencies nobody knew about - and it is the one piece of internal control that cannot be ticked off without evidence.

Insurers writing cyber insurance increasingly ask whether backups are kept off the network and whether a documented restore procedure exists. If after an incident the answers in the proposal form turn out to have been untrue, the claim may be reduced or refused.