Microsoft 365 Governance: Who Creates Teams, Who Invites, Who Administers
16 November 2026Browser Isolation with Zscaler: Keeping Web Threats Away from Your Devices
17 November 2026A backup that was never restored is an assumption. The Canadian Centre for Cyber Security asks organizations to "have
clear procedures on how to restore from back-ups and regularly verify that back-up and restore mechanisms operate as
expected"
(baseline cyber security controls, 3.7).
Get Cyber Safe puts it more simply: test your backups "to ensure they still open"
(Get Cyber Safe).
Here is how to go about it.
Two numbers to set before the first test
NIST defines two objectives:
- the recovery time objective (RTO): how long a system can be in recovery "before negatively impacting the
organization’s mission or mission/business processes"
(NIST, RTO); - the recovery point objective (RPO): "the point in time to which data must be recovered after an outage", in other
words how much work you accept to lose
(NIST, RPO).
Both numbers are decisions of the organization, not of the vendor. The test tells you whether they hold.
Restore aside, not on top
The Canadian Centre’s ransomware playbook says to "restore your systems into a clean, network-isolated location"
(ITSM.00.099). For a test, that is the rule too:
restore into a separate folder, machine or network, never on top of the data in use. CISA asks organizations to
"regularly test the availability and integrity of backups in a disaster recovery scenario"
(#StopRansomware Guide).
What to check, and what to write down
A useful test answers four questions:
- Does the file, mailbox or server come back whole, and does it open?
- How long did the restore take, compared with the recovery time objective? NIST notes that "an individual may
lack the tools or training to restore a particular system within the expected time period"
(SP 800-184). - How old is the restored data, compared with the recovery point objective?
- What was missing: a password nobody could find, a forgotten step, a licence?
NIST asks that the results of each exercise be documented in an after-action report
(SP 800-34), and that "the results
should be recorded to help inform organizational cybersecurity activities"
(SP 800-184). One line per test is enough: the date, what was restored,
the time, the result, what was fixed.
How often
The Canadian Centre’s ransomware playbook says: "You should implement a schedule to test your backups on a regular
basis (for example, monthly)" (ITSM.00.099). NIST
adds that exercises are repeated "following organizational or system changes"
(SP 800-34): a new server, a
migration, new backup software.
A possible calendar, to adapt:
- every month, a file or folder picked at random, and one mailbox;
- every quarter, a whole server or virtual machine, in a separate network, timed;
- after every major change, the system it touched.
Offline copies
The Canadian Centre says: "You should have 2 or more backups stored offline and inaccessible by your networks and
internet connection" (ITSM.00.099). "Because these
offline backups are not connected, they remain unaffected by many cyber threats, like ransomware"
(ITSAP.40.002). The test should reach
them too: an offline copy nobody knows how to reconnect does not help on the day of an incident. Our article on
the 3-2-1 backup rule explains how to spread them, and the one on
the first hour of a ransomware attack says what to do before
restoring.
Microsoft 365 too
Exchange Online keeps deleted items "for 14 days, by default", up to "a maximum of 30 days"
(Microsoft Learn).
Beyond that, only a backup kept elsewhere brings the data back, and it is tested like the others. For its Microsoft
365 Backup service, Microsoft says: "You should limit restores for testing purposes to no more than twice a month per
protection unit" (Microsoft Learn).
When the tool does part of the test
Some backup software automates part of the check. At Veeam, "SureBackup is the Veeam technology that allows you to test
machines backups and check if you can recover data from them"; it starts the machines in an isolated environment, runs
heartbeat, ping and application tests, then "creates a report on their state"
(Veeam Help Center). Our
Veeam page presents what we install. An automatic test does not replace a
person following the written procedure: that is who finds the missing password.
Frequently asked questions
How often should a restore be tested?
The Canadian Centre suggests a regular schedule, for example monthly. NIST adds that the test is repeated after a
major change in the organization or the system.
Where do you restore during a test?
Aside: into an isolated folder, machine or network, never on top of the data in use. The Canadian Centre says to
restore systems into a clean, network-isolated location.
What should be written down after a test?
The date, what was restored, the time taken compared with the recovery time objective, the age of the data compared
with the recovery point objective, the result and what was fixed. NIST asks that the results of each exercise be
documented.
What now?
MPJM’s monthly block of hours reserves the time to run these tests each
month, and the monthly security review keeps their result in writing. To
start, the free network analysis looks at where your backups are and
whether they restore. It is offered to organizations with 40 or more employees.
P.S. Sixteen questions, eight domains, a clear picture of your strengths and blind spots: start the assessment.
