Area 08 · Infrastructure and support

Backup

An archive becomes an archive not when it is created but when something is successfully restored from it. We configure backups with recovery in mind, test it regularly and know in advance how many hours it takes to get back to normal.

3-2-1
the rule of three copies
Monthly
a test restore
Off site
a second storage location
The plan
for an outage

What the area consists of

Three services answer three questions: what to back up with, where to keep it and what to do on the day everything fails.

Discuss the scheme

The two numbers everything depends on

Before choosing a tool you need two numbers for each system: how many hours it may be down and how much data loss you can live with.

Accounting databases

Downtime up to four hours, data loss within an hour. A single nightly backup is not enough; frequent transaction log snapshots are needed.

Shared folders

Downtime up to a day, data loss within a day. A single overnight backup outside working hours is enough.

Work email

Downtime up to four hours, data loss within an hour. The database backup is supplemented by a separate mail archive.

The website and the shop

Downtime no more than an hour, data loss within an hour. Meeting that target is impossible without a prepared standby site.

Staff computers

Downtime up to a day, data loss within a day. Achievable only if the documents are not kept on the workstation itself.

Something saved on the same hardware does not count as a backup. A failed drive, malware or an accidental deletion will destroy it along with the original. The working minimum is two copies on different media and one outside the building.

How the setup goes

The setup itself takes a few days, but the value appears at the fourth stage, which is the one most often skipped.

01

We find what needs backing up

We establish where the valuable data actually lives. We regularly find accounting databases on employees' personal machines and files that exist only as email attachments.

02

We work out the targets

Together we fix the maximum downtime and acceptable loss for each system. Those two numbers are what set the cost of the solution.

03

We configure and store

The frequency, the retention, encryption and a secondary site. We configure alerts for both outcomes: success and failure.

04

We verify by restoring

Every month we restore the backup on a separate rig and time it. There is simply no other proof that it works.

Questions and answers

The point is not whether you have backups but when you last restored from one. In our experience a fair share of supposedly configured schemes do not work: the job has been failing for months, the wrong folder is being saved, the file is locked with a forgotten password. Such a check takes a day and costs little.

Only if it cannot be reached from the same network with the same credentials. Modern malware deliberately hunts for backups. That is why you need a copy on a separate site, with its own permissions and immutability enabled.

The typical scheme: daily copies for two weeks, weekly for about two months, monthly for a year. That depth helps not only when hardware fails but also with a mistake discovered weeks later. For the finance department the periods are usually set longer.

It depends on the size of the data and the bandwidth. A fifty-gigabyte database restores from a local drive in an hour and a half, and noticeably longer from a remote site. We get the exact figure during a test restore and write it into the procedure so you know it in advance.

We will check whether your backups restore

Изучим действующую схему, попробуем восстановиться из имеющегося архива и без обиняков сообщим, выручит он вас при отказе или окажется бесполезным.