Dev.to Security ๐Ÿ” Cybersecurity ๐Ÿ‘ 0 ๐Ÿ“– 1 min read

Great SOCs run on systems, not heroes: measure Dwell Time, not counts

This is a read based on a fictional scenario; figures are illustrative. One great analyst can solve one case. But alerts arrive by the hundreds, people rotate, attacks evolve. Operations that depend on individual effor

This is a read based on a fictional scenario; figures are illustrative.

One great analyst can solve one case. But alerts arrive by the hundreds, people rotate, attacks evolve. Operations that depend on individual effort burn out.

What you'll learn

  • SOC tiers and two loops
  • Vanity metrics vs. outcome metrics
  • Handover and escalation essentials
  • Why you stop fatigue first, then get proactive

The story at a glance

02:20 root success to 02:41 C2 alert: about 21 minutes (one MTTD example). Log wiping after detection: a collection-design failure. A sloppy handover lets overnight activity surface in the morning, lengthening dwell time.

Tiers

Tier 1 triage and FP removal; Tier 2 deep investigation and containment; Tier 3 detection engineering and hunting; manager for KPIs and escalation. Tiers are roles, not rank. Small teams combine them but must keep every role.

Two loops

Reaction: logs -> detection -> alert -> triage -> response -> report. Improvement: lessons -> detection engineering -> hunting <- threat intel. The improvement loop makes the reaction loop lighter.

Metrics

Dwell Time (the key outcome), MTTD, MTTA, MTTR, FP rate, ATT&CK coverage and queue depth (a leading fatigue indicator). "Alerts handled" is a vanity metric: noise raises it too.

Handover and escalation

Handover notes: open cases and owners, watched events and held alerts, ongoing incidents, environment changes. Decide the escalation matrix in peacetime (Low to Critical, who approves isolation or account disablement).

Order of operations

  1. Stop the fatigue (cut FPs, automate tuned detections only).
  2. Invest the freed time in detection engineering, hunting, improvement.
  3. Feed findings back as detections.

Season one wraps here; next: building a detection lab and walking through a full exercise.

Adapted from chapter 37 of my Japanese book on server defense.

๐Ÿ“ฐ Read the original article on Dev.to Security

Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes โ€” full credit and traffic to the original publisher.