Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

The first hour of a suspected security incident on a hosted server

Technical article
Troubleshooting

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

In the first hour, do five things and avoid several others. Do: write down what you see, call the provider's security contact and your incident responder, isolate the affected system in a coordinated way, preserve evidence where that does not delay urgent containment, and check whether any reporting clock has started. Avoid: deleting or rebuilding the server, restoring over it, and using the possibly compromised email or chat to coordinate.

The hardest question is whether to power the system off. Two reputable government guides answer it differently, with different emphasis. For a server hosted by someone else, our reading is that you should not power it off or delete it on your own: you ask the provider to isolate it, and to preserve a snapshot and logs, and you record who agreed to what. No source we read defines a first hour; that framing is ours.

Why it matters

Early actions shape everything that follows. Evidence is lost when systems are wiped. An attacker who sees a clumsy response may move to other systems. A well-meant reboot can erase information held only in memory. A rebuilt server may be compromised again if nobody learned how the attacker got in; the CISA guide warns that a ransomware event may be evidence of an earlier, unresolved compromise.

On a hosted service there is a further complication: the customer and the provider each hold part of the picture. The provider can see the platform; the customer knows the application and the data. Neither can do the whole job alone, and the contract usually decides who may act.

Who does what: the provider and you

NIST describes this as a shared arrangement. Its incident response guide says third parties such as cloud providers may be involved in particular incidents, that the responsibilities should be clearly defined in a contract, and that the response team should know the division of responsibilities, including "information flows, coordination, and authority to act". It gives an example of a restriction on what a provider can do: making operational decisions such as "immediately deactivating certain services to contain an incident". NIST also notes that a provider may detect malicious activity sooner than a single customer, because it can correlate events across its customers.

Read that as a prompt to find out before an incident. Can the provider isolate a single virtual server from the network without your approval? Can it take a snapshot on request? Who at the provider can you phone at 2 am?

Two guides, two answers on powering off

The Australian Cyber Security Centre (ACSC) page on ransomware is aimed at a broad audience; its own tags list individuals and families, small and medium business, and large organisations, and its steps assume a device you can reach. After you record details of the attack, it says to turn off the infected device: "Hold down the power button or unplug the device from the wall". It also tells you to disconnect other devices and change important passwords as soon as possible.

The #StopRansomware Guide from US agencies is a longer checklist that, in our reading, assumes an environment with responders, and it orders the steps differently. It says to determine which systems are affected and isolate them, and it lists "Power down devices if you are unable to disconnect them from the network" as the second step. Its note explains the cost: powering down loses "ransomware infection artifacts and potential evidence stored in volatile memory", so it should be done only if it is not possible to take the network offline or disconnect the hosts by other means. For cloud resources, the same guide says to "take a snapshot of volumes to get a point in time copy" for later forensic investigation. Later, where no initial mitigation seems possible, it advises taking a system image and memory capture of a sample of affected devices, including cloud servers.

These are not contradictions so much as different priorities: stop the spread quickly (ACSC) versus keep evidence while still containing (CISA). Why the emphasis differs is our reading, not something either guide says. The CISA guide does cover cloud resources, for example taking a snapshot of volumes, but neither guide settles the contract and authority split between a customer and the provider that runs the server; that split is what you need to find out in advance.

Our reading for a hosted server: containment by network isolation, carried out or approved by the provider, is closer to the second guide. Power-off is a last resort. Evidence is preserved where that is feasible without delaying urgent containment: if isolation is not possible and shutdown is needed to stop the spread, a snapshot is not a condition to wait for, and the person leading the response weighs the trade-off. Where there is time, ask the provider to capture a snapshot before any shutdown. That decision belongs to the person leading the response, not to whoever notices first.

What to do in the first hour

This sequence is our recommendation, built from the sources above. Timings are a planning aid, not a standard.

First minutes: stop and record. Do not start fixing. Write down what you saw, when (with time zone), on which system, and who noticed. ACSC suggests recording when it happened, the ransom message, the device, any new file extensions, anything received from or shared with the attacker, and the steps you took, and notes that a photo of the screen is a quick way to capture it. Screenshots are on ASD's list of commonly collected evidence too.

Then: get the right people. Phone your provider's security or emergency contact, your internal decision-maker and your incident responder if you have one. ACSC says that ransomware attacks are hard to tackle alone and suggests finding an IT professional. Use the phone or another channel that does not depend on the affected systems. The CISA guide says to isolate in a coordinated manner and to use out-of-band communication such as phone calls, "to avoid tipping off actors that they have been discovered", because attackers may watch email and chat.

Then: contain with the provider. Ask for the affected server to be isolated from the network and from other systems. ASD's guidance describes the purpose of containment as limiting damage, stopping spread and "prevent malicious actors from destroying evidence". Ask what else the isolation will affect: other services that depend on the server may stop.

Then: preserve. Ask the provider to keep a snapshot or image, and the logs it holds, and to confirm in writing what it has preserved and for how long. A snapshot taken as evidence is not a backup: it is a point-in-time copy for investigators, kept apart, and it is not a restore point you should recover from. In our understanding a disk snapshot is also not a memory capture; the CISA guide lists the system image and memory capture as separate actions, so ask the provider which of them it can take. ASD's guidance lists commonly collected evidence: host images, network captures, log files, configuration files, databases, investigation notes and screenshots. It asks for a log recording "who collected or handled the evidence", with dates and times including time zone. It also says that, because evidence handling can be complex, you may need advice from digital forensic professionals, legal advisors or law enforcement.

Then: check clocks. ACSC says you may need to make some reports right away to meet legal, customer or insurance obligations. Check your insurance policy's notification wording and your contracts. If personal information may be involved, a separate assessment clock may start; see our companion article on notifiable data breaches.

What to avoid

  • Deleting or rebuilding the server. You lose the evidence, and you may not learn how the attacker entered. Rebuild after the cause is understood.
  • Restoring over the affected system. The CISA guide warns that a ransomware event may be evidence of an earlier unresolved compromise, so care is needed before rebuilding from backups. Recovery decisions come later; our companion article on recovering after ransomware covers that stage.
  • Coordinating on the compromised channel. Email, chat or remote sessions on affected systems may be visible to the attacker.
  • Changing credentials in a hurry from an affected machine. ACSC advises changing important passwords because some attacks steal them. The CISA guide's concern is that visible activity can prompt an attacker to move. Our reconciliation: do both, in a sequence led by the responder, from a clean device.
  • Paying a ransom as a first reaction. ACSC says never to pay a ransom; the CISA guide says its authors do not recommend paying, and notes that payment does not ensure data is decrypted.

Reporting to the government

ASD's reporting page says you can use it to "Report a cybercrime, incident, vulnerability or ransomware payment"; the hotline is 1300 CYBER1 (1300 292 371). In an emergency, ACSC says to call 000 where there is an immediate threat to life or risk of harm. ASD's guidance also describes a limited use obligation: information voluntarily provided to ASD about incidents cannot be used for regulatory purposes. Reporting to ASD does not replace any notification you owe to a regulator, a customer or an insurer.

Questions to put to your provider

  • Who is our 24-hour security contact, and how do we reach them without email?
  • Can you isolate a single server from the network on our request, and how fast?
  • Will you take a snapshot or image where that does not delay isolation, and who authorises that and the isolation itself?
  • Which logs do you hold for our account, and for how long?
  • Will you tell us if you detect suspicious activity affecting us, and within what time?
  • What do you need from us in writing to act?

A worked example (illustrative)

The following is invented. A small firm's file server shows renamed files and a ransom note at 7:40 on a Tuesday. The office manager photographs the screen, writes down the time and the affected share, and does not touch the server. They phone the IT contractor, who phones the provider's emergency number.

The provider isolates the virtual server from the network and takes a snapshot. The contractor asks for the server's logs to be kept and for written confirmation. The firm does not use its email to discuss the incident until the contractor has checked that the mail system is clean. By 8:30 the firm has a short log of what was done and who agreed to it, has told its insurer, and has submitted a report to ASD.

Risks and limits

  • No single source covers a customer in this situation. We joined a general Australian guide, a US ransomware guide and a NIST profile; the sequence is our recommendation.
  • Advice differs by device type, incident type and what responders are available.
  • The ACSC ransomware page was last reviewed in 2023; check for newer guidance.
  • Not every incident is ransomware. A suspected account takeover, for instance, calls for different first steps.
  • This article does not replace advice from a qualified incident responder. This is general information, not legal advice.

Next step

Write down now, before an incident, a one-page list: the phone numbers (not email) of your provider's emergency contact, your responder, your insurer and your decision-maker, and a note of who may authorise isolating or shutting down each system. Add a line on what evidence your provider can preserve. If you are reviewing a hosted service, the Managed Services page is a starting point for that conversation; ask for the scope in writing.

Sources

All accessed 15 November 2026.

Learning notes

  • Network isolation and powering off are different actions; power-off can destroy evidence held in memory.
  • Government guides can differ in emphasis; read who each is aimed at, and ask what applies to a server you do not hold.
  • On a hosted server, the provider holds the levers for isolation and snapshots; agree who may pull them before you need to.
  • Write down what you see before you change anything, with the time zone.