This is a concept website · Enquire about this domain

dedicated-servers.com.auThe stack

Guide 2 of 748 h

How fast does a server have to be patched?

ASD sets patch deadlines layer by layer, and the shortest starts the moment a fix is released, when the vendor rates the flaw critical or a working exploit exists. This page follows a single release from the floor of the machine to the data sitting on it.

General information, not legal or security advice.

A release, followed up the stack

The worked example: one release, and the deadline the ISM sets for each layer
  1. Firmware and drivers take the short clock when the flaw is serious: patches, updates or other vendor mitigations for vulnerabilities in firmware and drivers are applied within 48 hours of release when vulnerabilities are assessed as critical by vendors or when working exploits exist. Where neither is true, both take one month.
  2. The operating system of an internet-facing server takes the same pair: 48 hours when the vulnerability is assessed as critical by the vendor or a working exploit exists, and two weeks otherwise.
  3. Online services take the pair again: 48 hours and two weeks, on the same conditions.
  4. Other applications, meaning anything that is not an office productivity suite, a web browser or its extensions, an email client, a PDF application or a security product, take one month. This layer has one clock only: the ISM sets no critical-flaw deadline against it.
The deadlines, as the ISM states them
LayerCritical, or working exploitOtherwise
Firmware48 hoursML31 monthML3
Drivers48 hoursML31 monthML3
Operating system (internet-facing server)48 hoursML1 to ML32 weeksML1 to ML3
Online services48 hoursML1 to ML32 weeksML1 to ML3
Other applicationsno critical clock set1 monthML1 to ML3

Source: ASD, Information Security Manual, Guidelines for system management, September 2026.

One month is not a fixed number of days, which is why the stack writes each deadline as type rather than drawing a bar for it: a bar length would assert a number of days the ISM never sets. Each figure is a deadline counted from release, and the clock in this guide's running head, 48 hours, is the shortest of them.

What critical and exploited mean

The short clock does not turn on how alarming the news sounds. It turns on two things ASD names in the control itself: the vendor has assessed the vulnerability as critical, or a working exploit exists. It also helps to be precise about the word vulnerability: in ASD's patching guidance a vulnerability is a flaw in an application or operating system rather than a misconfiguration, so a hardening checklist and a patch clock are different instruments, and a server needs both.

The clock starts when the patch is released. When patches or updates are released by vendors for vulnerabilities, an organisation should apply them in a timeframe commensurate with the likelihood of attempted exploitation by malicious actors. Once a patch is released, an organisation should apply it in a timeframe commensurate with its own exposure to the vulnerability. ASD puts a number on the exposure for an online service: once the vulnerability is public, malicious code can be expected within 48 hours, sometimes within 24.

Where no patch exists yet, ASD does not leave the gap empty: temporary workarounds may provide the only effective protection if vendors have no patches for the vulnerabilities. And whatever the route, ASD's standing line is that security patches should be applied as soon as possible, through a centralised and managed approach that maintains the integrity of the patches and confirms they were applied successfully, with a software register keeping the versions and patch histories of applications, drivers, operating systems and firmware.

Scanning, and the 48-hour expectation

The deadlines are only as good as the machinery that finds what needs patching. ASD's guidance expects a process to identify, assess and apply security patches as soon as possible, with an inventory of applications and operating systems to catch what is missing, and, where resources are tight, patches applied first to internet-facing infrastructure. Where everything cannot be done at once, ASD says to prioritise the vulnerabilities that pose the greatest risk to systems and business operations. At Maturity Level One of the Essential Eight, the model's own control is specific: a vulnerability scanner is used at least daily to identify missing patches or updates for vulnerabilities in online services.

Scanning is the finding half; the checking half matters just as much. ASD asks that patches be regularly confirmed as successfully applied, and that staff know their responsibilities and that security warnings from software vendors are monitored. That regular check is what catches a patch that silently failed to install.

Patching through a change freeze

A change freeze is a pause on changing systems, usually for a busy trading period, and ASD treats the patch clock as one of the things a freeze should not stop. Organisations should ensure that vulnerabilities are still being addressed during change freeze periods, especially within 48 hours for internet-facing infrastructure. Critical vulnerabilities, and vulnerabilities that affect critical applications or operating systems, should also be addressed with patches or other recommended mitigations from vendors while the freeze is on.

The trade-off is real, and ASD has weighed it: the immediate protection from patching a vulnerability that is being exploited right now far outweighs the impact of the unlikely event of having to roll the patch back. The preparation that makes this safe is unglamorous: clear patch management processes and procedures that account for the possibility of faults during patching. On a single server the rehearsal is small: a change window is a reboot, and the rollback plan can be a known image.

When a layer reaches end of support

Every clock above assumes somebody is still releasing fixes. When a vendor stops, the ISM's answer changes from a deadline to a replacement: operating systems that are no longer supported by vendors are replaced, online services that are no longer supported are removed, and networked IT equipment that is no longer supported is replaced. ASD explains the reason directly: as applications, operating systems and equipment reach their cessation date for support and become legacy IT, an organisation finds it increasingly difficult to protect them, because patches, updates and other forms of support are no longer made available by vendors.

Replacement is not always possible on the day, and the ISM says what to do then: when unsupported items cannot be immediately removed or replaced, compensating security controls are implemented until such time as they can be removed or replaced.