Key takeaways
- A backup you have never restored is a hope, not a backup. Test restores on a schedule.
- Follow the 3-2-1 rule: three copies, on two kinds of storage, one of them off-site. Add one copy that cannot be changed or deleted.
- Cloud services such as Microsoft 365 keep the service running; protecting your data from deletion and attacks is still your job.
- Decide how much data you can lose and how long you can be down first, then choose the technology.
Why backups fail when it matters
Backups rarely fail at the moment you make them. They fail later, and nobody notices until the day they are needed. Three causes come up again and again.
The backup job has been failing silently for weeks. The backup sits on the same network and the same credentials as the production systems, so ransomware encrypts it too. Or the data is there, but nobody knows the order in which to restore systems or how long it would take.
The 3-2-1 rule, and the extra copy
The classic rule is simple enough to remember and strong enough to matter.
- 3 copies of your data: the live data plus two backups.
- 2 kinds of storage, for example a local disk or appliance and a cloud object store.
- 1 copy off-site, so a fire, theft or flood at the office does not take everything.
- 1 copy that cannot be changed or deleted for a set time (immutable storage) or that is disconnected from the network (offline). This is what survives a ransomware attack that steals administrator rights.
Decide your recovery targets first
Two numbers drive every backup design. The recovery point objective (RPO) is how much recent data you can afford to lose: a day, an hour, five minutes. The recovery time objective (RTO) is how long a system can be down before the business is hurt.
Not everything needs the same answer. Sort your systems into tiers, for example critical (hours), important (a working day) and the rest (several days). The tiers tell you where to spend money and in which order to restore.
Software-as-a-service data needs a backup too
Providers such as Microsoft and Google run the platform and keep it available. That is not the same as a backup of your mailboxes, files and teams. Deleted items stay in a recycle bin for a limited time, and a compromised account or a malicious insider can delete or overwrite a lot quickly.
An independent backup of your cloud data, stored outside the provider’s own tenant, closes that gap. Check the retention settings of the service as well, and read what your provider says about its own responsibility.
Test restores on a schedule
A restore test turns an assumption into evidence. It does not need to be large.
- Every month, restore a few files and one mailbox item and check that they open.
- Every quarter, restore a whole server or virtual machine into an isolated test network and note how long it took.
- Once a year, rehearse a bigger scenario: your main systems are gone, who does what, in which order?
- Write the results down. If the real time exceeds your RTO, you have found a gap while it is still cheap to fix.
- Make failed backup jobs raise an alert that a person actually reads.
Common mistakes to avoid
A few habits quietly undermine otherwise good setups.
- Using the same administrator account for the backup system and for the whole network.
- Keeping the only copy of the encryption keys on the system that is backed up.
- Backing up servers but not laptops, SaaS data or the configuration of the network equipment.
- Having no written restore procedure, so recovery depends on one person’s memory.
Monitored backups and scheduled restore tests can be part of managed services, and they fit naturally with the protective measures described in our security services.