← all systems · see it in The System
Security operations center (SOC)
A unit responsible for protecting an organization against threats (src:wp:security-operations-center); in information technology a dedicated site where enterprise information systems are monitored, assessed and defended, where analysts monitor the network, investigate potential incidents and take steps to remediate them (src:wp:information-security-operations-center). Researchers define it as 'an organizational aspect of an enterprise's security strategy' that 'combines processes, technologies, and people to manage and enhance an organization's overall security posture' (src:vielberth-2020-soc). Walked as the team, the room, the tools and the routines: tiered analysts and managers, the SIEM and its alerts, triage, investigation and escalation, playbooks and metrics, shifts, and the burnout and alarm fatigue the research describes.
Properties published scienceour proposal
Sourced fields, and our own readings.
| aliases | Security operations centres (analysts) |
| born at | about 2005 ('around 15 years ago' in a 2020 review); forerunners 1962 (AT&T Network Control Center), 1968 (NSA SIGINT watch centre), 1988 (first CERT) |
| born precision | years to a decade |
| boundary | Inside: the SOC's people (tier 1 triage analysts, tier 2 incident responders, tier 3 threat hunters, SOC managers, with security engineers, threat intelligence and malware analysts around them), its room and shifts, its tools (the SIEM as the 'single pane of glass', ticketing, automation and SOAR, dashboards), and its processes (data collection, detection and analysis, triage, escalation, containment, playbooks, metrics, governance and compliance). Outside: the organisation's systems it watches (networks, hosts, applications, endpoints) and the employees who use them; the detection tools that feed it (IDS, antivirus, firewalls, which other walks own); the attackers; the organisation's management, regulators and auditors; outsourced providers (MSSPs) and the network operations centre it works beside. The edge with incident response is contested (a CSIRT is 'classified either as an independent team or part of a SOC': C-1588). |
| nested level | An organisational unit (rung 12) inside an organisation, the operating heart of its information security; built on the organisation's computer networks and on detection technology; staffed by people whose vigilance, judgement and morale are its upper floors. |
| scale | social |
Measures (21) published science
| characteristic clock: staffing cycle | 8x5 to 24x7 hours a day x days a week |
| shift pattern (corporate SOC) | L1: 4 days x 10 h; L2: 5 days x 8 h shifts |
| characteristic clock: discovery peak in the working day | 8-10 AM time of day |
| vigilance decrement onset | within 15 (as little as 8) minutes of sustained attention |
| analyst tenure in the job | 1-3 years (HP white paper, cited) |
| alerts per day (NCSA SOC, 2018-2022) | 24,000-134,000 alerts per day |
| share of alerts associated with true attacks | 0.01 % |
| share of alerts that are benign triggers | 49 (attack attempts 27) % |
| share of alerts auto-blocked (BHR) | 44.9 % (6,474,204 alerts, 2020-2022) |
| alerts per day reaching analysts after auto-block | about 13,764 alerts per day |
| true attacks detected the same day; post-attack investigation | 66% same day; 53 days average % ; days |
| mean time to detect a breach (industry report, 2018) | 196 (plus 69 to contain) days |
| share of alarms analysts call false positives (perceived) | 99 % (one analyst's figure) |
| share reporting low network visibility | 71.43 of analysts, 60 of managers % |
| mitigation deadline by priority (one SOC) | P0 30 min, P1 60 min, P2 120 min, P3 24 h, P4 48 h; 87% met time to mitigate |
| threat intelligence volume held (one SOC) | 17 to 20 terabytes in two months |
| devices watched (corporate SOC) | about 300,000 devices online a day |
| devices watched (university SOC) | 55,000 devices in business hours (4 analysts) |
| time to correlate logs and file one ticket (manual) | 10 (about 15 such alerts a day) minutes |
| cybersecurity workforce gap (worldwide) | roughly 4 million people |
| practitioners saying false positives are not a major issue | 15 of 18 interviewees |
Open contentions (5) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: subconscious pattern recognition from experience (skilled intuition, Kahneman and Klein); other theories of where thoughts come from (see gameplans/theories-of-mind.md) - C-1496 Are 99% of security alarms really false?
booked: The map books the analysts' own count, and keeps the finding that most such alarms are real events on harmless activity.
the other reading: most alarms are true alarms on legitimate activity; false alarms in the strict sense dominate - C-1572 Are false alarms a big problem in security operations centres?
booked: The map keeps both: analysts who call it overwhelming, and practitioners who say it is not a major issue.
the other reading: false positives are the SOC's central burden; false positives are routine and tuned away; low visibility is the real problem - C-1588 Is incident response part of the security operations centre?
booked: The map books it as a separate team the centre hands off to, and keeps the reading that it sits inside.
the other reading: the CSIRT/incident response team is part of the SOC; the SOC is only the monitoring unit; response belongs to incident response - C-1589 When did security operations centres begin?
booked: The map keeps the candidates: around 2005, the 1960s watch centres, or growing out of response teams.
the other reading: the SOC is the latest kind of one older system, the watch centre (signals intelligence, network control, physical security operations), born in the 1960s or earlier; the SOC grew out of CSIRTs (1988 on), as the encyclopedia says CSIRTs evolve into ISOCs
Partners and parts published science
Built from / builds (11)
- Computer security
- Security information and event management (SIEM)
- Cyber threat intelligence
- SOC analysts (tier 1, tier 2, tier 3)
- SOAR and automation
- SOC manager
- Playbooks and standard operating procedures
- SOC metrics and reports
- Attackers against defenders for control of computing
- Organisations and providers competing for skilled analysts
- Competition for undetected progress (red team vs blue team)
Sits inside / holds (5)
- Intrusion detection system
- Computer security
- Information security
- Organization
- Security information and event management (SIEM)
Trades with (valence) (19)
- Intrusion detection system via sys:electromagnetic-field · Security operations center (SOC) mixed
- Human (Homo sapiens) via sys:sensory-environment · Security operations center (SOC) gain
- Computer security via sys:human-society · Security operations center (SOC) gain
- Malware via sys:electromagnetic-field · Security operations center (SOC) gain
- Incident response via sys:electromagnetic-field · Security operations center (SOC) gain
- Threat actor via sys:computer-network · Security operations center (SOC) mixed
- Organization via sys:human-society · Security operations center (SOC) mixed
- Computer network via sys:electromagnetic-field · Security operations center (SOC) gain
- Security information and event management (SIEM) via sys:written-record · Security operations center (SOC) gain
- Cyber threat intelligence via sys:written-record · Security operations center (SOC) mixed
- Outsourced incident response (MSSP, contractors) via sys:market · Security operations center (SOC) mixed
- Network operations center (NOC) via sys:human-society · Security operations center (SOC) gain
- Phishing via sys:internet · Security operations center (SOC) gain
- Risk management via sys:organization · Security operations center (SOC) gain
- Insider threat via sys:organization · Security operations center (SOC) gain
- Vulnerability management via sys:organization · Security operations center (SOC) gain
- Social engineering via sys:human-society · Security operations center (SOC) gain
- Penetration testing via sys:computer-network · Security operations center (SOC) mixed
- Security awareness via sys:written-record · Security operations center (SOC) gain
expressed by (25)
- Generation/Synthesis
- Templating / Standardization
- Quality Improvement/Refinement
- Interoperability/Transduction (Bridging)
- Symbiosis/Co-Creation (Collaborative Specialization)
- Distributed Autonomy/Decentralized Execution (Subsidiarity)
- Internal Coordination/Regulation
- Sensing/Scanning
- Mapping/Modeling
- Uncertainty Reduction/Learning
- Multi-Scale Data Fusion / Triangulation
- Trail-Marking / Externalized Memory (Stigmergy)
- Surprise / Change-Point Detection
- Frontier Outposts / Sensing Extension
- Offline Simulation & Replay-Guided Planning
- Social Learning / Public Information Use
- Quick Harm Avoidance
- Reactive Response
- Proactive Defense
- Technique-Based Identification (a.k.a. Technique-Based Detection)
- Self-Monitoring
- Validation of Effectiveness
- Redundancy & Diversity
- Triage & Prioritized Allocation
- Quality Control & Error Correction
floor filled by (20)
- Computer security
- Intrusion detection system
- Human (Homo sapiens)
- Incident response
- Threat actor
- Organization
- Firewall (computing)
- Computer network
- Security information and event management (SIEM)
- Black hole router (automatic blocking)
- Chief information security officer
- Antivirus / anti-malware software
- Cyber threat intelligence
- SOAR and automation
- SOC manager
- Playbooks and standard operating procedures
- Tier 1 analyst (triage specialist)
- Tier 2 analyst (incident responder)
- Tier 3 analyst (threat hunter)
- Information security
Kinds (8)
- Outsourced incident response (MSSP, contractors)
- In-house (internal) SOC
- Hybrid SOC
- Virtual SOC
- Tiered SOC
- National SOC
- Distributed and decentralized SOC
- Physical security operations centre
wired (6)
- Human (Homo sapiens)
- Intrusion detection system
- Incident response
- Consoles, dashboards, search node and reporting
- Threat intelligence analysts
- Reporting button and crowd-sourced reports
Notes published science
- A SOC finds attacks on office hours: at one large SOC more attacks started between 6 PM and 8 AM, but most were discovered between 8 and 10 in the morning, when the analysts were in (src:yang-2024-soc-alerts).
- Of 24,000 to 134,000 alerts a day at that SOC, about 1 in 10,000 was tied to a real attack, and almost half were "benign triggers": true alarms on things the organisation had decided to live with (src:yang-2024-soc-alerts). Intensive-care monitors have the same problem: 72% to 99% of their alarms are false or not actionable (src:chromik-2022-icu-alarm-fatigue).
- Burnout runs as a loop: low skills mean managers withhold trust, which kills creativity and growth, which keeps skills low; the same loop runs the other way once managers take a small risk and empower analysts (src:sundaramurthy-2015-burnout). The researchers found it by having a student work as an analyst for six months, the way anthropologists live with a village (src:sundaramurthy-2015-burnout).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
the boardroom (shared floors)
- L5: Tier 1 triage: analysts at the SIEM console validate each alarm (true, false, benign trigger), enrich it, prioritize it and escalate what they cannot solve; their vigilance, tacit knowledge, hunches and morale bound what is seen
- L6: Investigation and response: tier 2 incident responders assess escalated incidents in depth with threat intelligence, scope the attack, contain it; tier 3 threat hunters look for the unknown; situational awareness of the estate
- L7: Management and governance: the SOC manager and CISO set what the SOC watches and how (strategy, playbooks, staffing, budget, metrics, compliance), aiming at the organisation's security posture
the security operations center as SOC researchers and analysts describe it: the organisation's systems it watches up to the security posture it serves (reader's view)
- L1: The watched estate: enterprise information systems (websites, applications, databases, servers, networks, endpoints), the people who use them, and the threat landscape outside
- L2: Sensors and sources: security software (IDS/IPS, firewalls, antivirus, vulnerability scanners), network assets, operating systems, physical security, threat intelligence, employees
- L3: Collection: logs and alerts gathered, normalized, time-synchronized, filtered, reduced, aggregated and prioritized
- L4: Correlation and alarming: the SIEM's correlation rules turn events into alarms on dashboards, the single pane of glass
- L5: Tier 1 triage: validate, enrich, prioritize, escalate
- L6: Tier 2 and 3: in-depth investigation with threat intelligence, containment, recovery, threat hunting
- L7: Management and governance: the security posture, regulation, budget, staffing, metrics
way in: Hunches and tacit knowledge (inner)
- L1: Tacit knowledge, experience and understanding of the monitored network
- L2: Hunch feelings, commonly accurate
- L3: Skilled intuition: quick, heuristic (Type 1) decisions, hard to explain
way in: Knowledge of the organisation's own network and devices (inventories, architecture, onboarded devices)
- L1: The organisation's network setup and devices, often only partly known to the SOC
- L2: Network architecture documentation, device lists, asset inventory, configuration management database
- L4: Onboarding: the devices the SOC agreed to receive events from bound its view
way in: Management, regulators and the organisation's demands (metrics, budget, compliance)
- L1: Reasons and rules: sensitive data, industry rules (PCI DSS), government rules
- L2: Metric pressure relayed down
- L3: Funding follows management's perception of value
way in: Reports from employees (human-as-a-security-sensor)
- L1: Employees' daily work and mail, where phishing is the attack most often met
- L2: An employee notices a phishing mail and reports it
way in: Security alerts from detection tools (IDS, antivirus, firewalls, endpoint tools)
- L1: Activity on the enterprise information systems: websites, applications, databases, servers, networks, desktops and other endpoints
- L2: Security tools and network devices raise alerts when they detect a threat (signature- or anomaly-based IDSs and others)
- L3: Alerts and events collected into the SIEM: normalized, filtered, reduced, aggregated
- L4: Correlation rules (directive alarms) and prioritization turn alerts into alarms on the analysts' dashboard
way in: Sound: the room's soundscape and sonified network data
- L1: Non-speech audio carrying data (sonification), and the room's own soundscape (a radio, talk, headphones)
- L2: Analysts' hearing: attacks can be heard in sonified traffic in some contexts; one SIEM has a sonification module
way in: Strain and morale: monotony, frustration, exhaustion (interoceptive)
- L1: Repetitive, monotonous triage under pressure
- L2: Felt as frustration and lost confidence
- L3: Shows in work: exhausted analysts stop giving detailed updates and cherry-pick events
way in: System, application, network and security logs
- L1: Logs record the state of each system and the changes in it
- L2: Log generators: operating systems, applications, switches, routers, servers, proxies, mail servers
- L3: Collected, pushed or pulled, normalized to one format and time zone, filtered, reduced and aggregated
- L4: The SIEM correlates log events, enriches them with context and alerts
way in: The SOC's own metrics (event volume, missed events, processing time, coverage)
- L1: The SOC's own work stream: events received, processed, left unprocessed
- L2: Metrics generated automatically by the SIEM
- L4: Compared across shifts and periods
- L5: L2 analysts turn them into metrics and reports for management
way in: Threat intelligence (feeds, open sources, sharing platforms, other organisations)
- L1: The threat landscape as others record it: open-source intelligence, sharing platforms, other organisations
- L2: Open-source and commercial feeds
- L3: Volume held and filtered (17 to 20 terabytes in two months at one SOC; half to be filtered out)
- L4: Context for analysis: indicators and reports enrich alerts
- L5: Dedicated threat intelligence teams and analysts
way out: Automatic blocking without analysts (Black Hole Router; a reflex set by policy)
- L1: The offending address is blocked on the network
- L4: A high-confidence alert triggers the block in real time, skipping the analysts
way out: Contain and remediate (block, deactivate, reimage)
- L1: The attack activity is cut off on the network or host
- L3: Orchestration and automation tools carry it out
- L6: Decision on the evidence: block an activity, deactivate an account, watch, or escalate
way out: Escalate to incident management / incident response
- L6: Incidents needing in-depth analysis handed to the incident management team
way out: Notify users and device owners (tickets, emails)
- L1: Read by the user (on Human's sight channel), who reimages the device
- L5: Remediation instructions written to the user
way out: Report to management (metrics and stories)
- L1: Read by upper management, who fund or close the SOC
- L7: The SOC communicates its value upward
way out: Tune the detection rules
- L2: The detection tools' rules are changed (the IDS's rule set)
- L5: False positives corrected one by one after validation
Who fills each layer
- L3: Security information and event management (SIEM)
- L4: Security information and event management (SIEM)
- L6: Cyber threat intelligence