Whoever may isolate at night must know it beforehand
At 2:40 a.m., endpoint detection reports ransomware behaviour on the file server. The on-call engineer could take it off the network immediately, but are they allowed to? NIST SP 800-61r3 says containment is faster when criteria and procedures are set in advance. What that means in practice: who may do what without asking, what leadership decides, and when an immediate action ends again.
2:40 a.m. Endpoint detection reports a pattern on the file server that looks like encryption: many files, new extensions, a process nobody knows. The on-call engineer has the console open and could take the server off the network with two clicks.
They do not do it immediately. The file server is critical, the early shift needs it at six, and nobody has ever said whether on-call may decide something like this alone. So they call the head of IT. Voicemail. Then the deputy. After forty minutes the decision is made. The encryption kept running all that time.
What NIST says
The incident response recommendations NIST SP 800-61 Revision 3 of April 2025 define containment briefly: "Containment refers to preventing the expansion of an incident." And they name the bottleneck from the scene: manually selecting containment and eradication actions may be "easier and faster if the organization has criteria and procedures in place".
Who decides is also addressed. Among the roles, NIST names leadership, which may have decision-making authority on high-impact response actions, "such as shutting down or rebuilding critical services". And for duration, the document suggests distinguishing at the time of selection: an emergency workaround that must be removed within hours, a temporary workaround for up to two weeks, or a permanent solution.
Automatic containment is explicitly foreseen, such as quarantining malware or moving a device into an isolated network. An organisation can also authorise third parties such as internet or cloud providers to contain certain incidents, such as large DDoS attacks, on its behalf. Both require that someone has decided beforehand what may happen automatically.
What should be settled before the incident
- Which immediate actions exist. Taking a device off the network, locking an account, ending sessions, blocking an address or domain. For each: with which tool, and how it is reversed.
- Who may do what without asking. On-call may probably always isolate a workstation. The domain controller or the control system of a production line, probably not. This line belongs in writing per system class, not asked for at night.
- When leadership decides. For actions that shut down critical services, and who deputises when leadership cannot be reached.
- How long an action applies. A block without an expiry date becomes a permanent rule nobody understands any more. Every immediate action needs an end or a deliberate transfer into a permanent solution.
How isidaten maps this
In isidaten's containment module, every action, whether host isolation, account lock or session termination, goes through a four-eyes approval and is logged on the incident. Reversal is a separate order with its own approval, and blocklist entries have an expiry date. Hosts and accounts that must never be contained automatically, such as domain controllers or emergency accounts, can be stored as protected targets. A manual request remains possible but requires explicit confirmation at approval.
The test question
May your on-call engineer take the file server off the network tonight without calling anyone? And is the answer written down where they can look it up at 2:40 a.m.?
Sources
NIST SP 800-61 Rev. 3, Incident Response Recommendations and Considerations for Cybersecurity Risk Management, April 2025, section 2 (roles) and table for CSF element RS.MI (Incident Mitigation), RS.MI-01.
Questions about this update?
Talk to us – we are happy to show you this feature in a demo.

