Guide 4 of 7Backed up by business criticality
Does the host back up a dedicated server, or do you?
On a rented machine a backup can be three different things: part of the service, an option you buy, or a process you have to run yourself. This page puts ASD's question list beside what the ISM says a reliable backup looks like, and settles the two into one sequence.
General information, not legal or security advice.
Whose job it is
Start with ASD's own account of the arrangement, because it names all three possibilities in one sentence: a backup process might be part of the service, it might be offered as an optional extra, or you might be required to implement a backup process yourself. The word that matters is "might", and it is the argument for reading the provider's responsibility document before the contract rather than after the first restore request.
These are the questions ASD lists for a customer to ask a provider. CSP, in ASD's wording, means a cloud service provider; on this site, read it as your hosting provider. Under backups, ASD's six questions are:
- Does the CSP back up your data or do you need to?
- Are backups automatic and free, or do you need to set them up and pay extra?
- Do backups occur often enough, and can you control their frequency?
- Do backups include all of your important data, and can you control what is included?
- Are backups protected from being accessed, changed or deleted without your approval?
- Do you need to store a copy of the backups elsewhere, and will that cost you extra?
Under restoring data, ASD's list turns from the copy to the recovery:
- Can you restore data yourself, or do you need help from the CSP?
- Can you restore earlier versions of your data or only the latest version?
- Have you tested restoring data from a backup?
Those nine answers decide which of ASD's three arrangements you are actually in. They also set the row of the stack this guide lights: your data carries no patch clock, because patching is the wrong instrument for it. What stands in the clock's place is a different rule, and the ISM puts it in one line: backups of data, applications and settings should be performed and retained in accordance with an organisation's business criticality and business continuity requirements. The cadence of your backups is set by what the business would lose if the data were gone, not by a vendor's release cycle.
What a good backup looks like at the ISM
The second document in the conversation is the Information Security Manual, which describes a backup that can be relied on rather than one that merely exists. Its parts fit together as a picture.
The copy is protected from the people who might want it gone: backups are stored using a technically enforced immutability mechanism that prevents their modification or deletion for the duration of their retention period, and unprivileged user accounts are prevented from modifying and deleting backups at all. The copy also lives apart from the machine it protects: backup infrastructure, including backup servers, repositories and management consoles, is segregated from production environments and uses a separate authentication mechanism for administrative access. A backup that sits on the same server, under the same administrator login, meets neither of those last two tests.
Then the part most often skipped. Restoration of data, applications and settings from backups to a common point in time is tested as part of disaster recovery exercises, which is the ISM's version of ASD's plainer advice to regularly back up important data and test your ability to restore it. The Essential Eight's Maturity Level One states the same three things as controls: backups of data, applications and settings are performed and retained in accordance with business criticality and business continuity requirements; restoration from backups to a common point in time is tested as part of disaster recovery exercises; and unprivileged user accounts cannot modify or delete backups. Having data backup and restoration processes and procedures is also, in ASD's words, an important part of business continuity and disaster recovery planning. The reason it matters is worth keeping in view: regular backups let a business recover its data if it is lost, damaged or encrypted by ransomware, and they are a key defence against physical events like fires or floods as much as against cyber incidents.
One sequence to settle it
Put the two documents together and the disagreement resolves into an order of operations. First, ask ASD's questions above, in writing, and get the answers in writing; the one that decides everything is the first: whose job is the backup? Second, read the answer against the ISM's picture: immutable for its retention period, on infrastructure separate from what it protects, under a separate administrator login, and restore-tested. Third, where the service leaves a gap, either close it yourself or ask the provider to close it in the contract, and put the date you checked beside it. The sequence works for a managed arrangement and for an unmanaged one; what changes is how many of the gaps are yours to close.
One worked line makes the third step concrete, and it is the smallest version of ASD's restore test. On the day you first try it, a restore proves that the backup can be read back and that someone in the business knows how to do it. It does not prove that the backup contains everything important, and it does not prove that a full restore finishes inside the time the business can be offline. Those two are the questions to put to the next test, and the ISM's common point in time is what a full one should show.
The whole stack, bottom to top (dedicated-servers.com.au home)