Topic 19 ยท Deep Dive

19.1 What an Incident Is and How to Classify It

Events Are Cheap; Incidents Are Expensive

An event is any observable occurrence on a system: a login, a failed handshake, a process spawn, a firewall drop. A security incident is a narrower thing: an event, or a chain of events, that actually threatens confidentiality, integrity, or availability (CIA). Thousands of events fire per hour; only the ones that endanger data or operations earn the incident-response process. The classification gate exists so the IR team never drowns in log noise — and so a genuine compromise never slides into the ticket queue as "just another alert."

The Severity Ladder

Most shops run a three-rung ladder. P3 (alert): suspicious but contained or unverified — normal operations handle it on shift, monitor, and log; no IR activation. P2 (incident): a confirmed CIA impact on a live system — on-call IR responders are paged and the containment clock starts. P1 (breach): confirmed exposure of regulated or sensitive data, or widespread outage with legal implications — legal, communications, and executives join, and notification windows under law or cyber-insurance contracts start ticking, sometimes measured in hours. The one design rule that matters: severity definitions must be written down before the 2 a.m. page, because improvised judgment under stress is how incidents get mis-served.

Where Classification Sits in the Lifecycle

NIST SP 800-61 splits the lifecycle into preparation; detection and analysis; containment, eradication, and recovery; and post-incident activity. Classification lives inside detection and analysis, right after validation: analysts filter noise, then assess impact from asset value and incident magnitude — quantitative estimates (currency lost or extractable) plus qualitative categories (data sensitivity, reputational reach). Misclassification costs in both directions: over-escalating false alarms burns responder capacity and trust; under-escalating a real breach burns the clock, which ransomware operators and notification deadlines spend happily.

Decide Fast, Document Anyway

Time-box the initial severity call, give one named role ownership of it, and record the rationale, the decider, and the time. Severity can climb or fall as facts mature; the audit trail of those moves becomes part of the incident record, and insurers and regulators read it.

Architecture Diagram

Classification gate and severity ladder Event CIA affected? validated impact P3 alert (ops) P2 incident (IR) P1 breach (legal) no / low impact confirmed regulated data
Every event passes one gate — does it threaten confidentiality, integrity, or availability? — and survivors climb a pre-agreed severity ladder.

Key Takeaways

« Back to Topic 1919.2 »