System Components

Chapter 1 — Core Switch Security Hardening Design Guide


1.1 System Architecture

The core switch security hardening system is composed of multiple interdependent zones, each with a defined trust boundary and set of responsibilities. Understanding the architecture and the relationships between zones is essential before implementing any hardening controls, as misalignment between zones is a common source of both security gaps and operational failures.

The architecture is divided into five primary zones: the Admin Zone (where operators and SOC/NOC personnel work), the Secure Ops Zone (containing jump hosts, bastion servers, AAA, config backup, and PKI), the Network Core Zone (the core switch pair and their upstream/downstream connections), the Out-of-Band Management Network (providing isolated management connectivity), and the Logging & Time Zone (Syslog/SIEM, telemetry collectors, and NTP). All management access must traverse the Secure Ops Zone before reaching the Network Core Zone, ensuring a consistent audit trail and enforcement point.

System Component and Boundary Architecture Diagram
Figure 1.1: System Component and Boundary Architecture — Five Security Zones with Trust Boundaries and Access Flows

The Out-of-Band Management Network is the most critical dependency for operational resilience. It provides management access even when the production network is degraded or under attack. The OOB network must be physically or logically separate from the production data network, with its own management switch, firewall, and IP addressing plan. All core switch management interfaces (mgmt0, mgmt1, or equivalent) must connect exclusively to the OOB network, never to production VLANs.

Module relationships follow a strict hierarchy: operators authenticate through the Admin Zone, pass through the Secure Ops Zone (jump host + AAA), traverse the management firewall, and reach the core switch management VRF. Logging and telemetry flow outbound from the core switch to the Logging & Time Zone. NTP synchronization flows inbound from the NTP hierarchy to the core switch. Config backups flow outbound to the backup repository in the Secure Ops Zone.

Core vs. Optional Components

Not all components carry equal weight in the hardening baseline. The following classification distinguishes mandatory core components from optional enhancements that improve posture but are not strictly required for baseline compliance.

CategoryComponentJustification
CoreManagement isolation (OOB or mgmt VRF)Prevents lateral movement from production to management plane
CoreAAA (TACACS+/RADIUS)Required for audit, RBAC, and command accounting
CoreSyslog/loggingRequired for forensics, compliance, and incident response
CoreNTP (restricted)Required for log correlation and certificate validity
CoreConfig backupRequired for rollback and change management
CoreCoPP/CPPRequired to prevent CPU exhaustion attacks
CoreProtocol hardening baselineRequired to prevent routing attacks and L2 spoofing
OptionalAutomated compliance scanningImproves posture verification speed and coverage
OptionalStreaming telemetry at scaleImproves observability and MTTR
OptionalSOAR for automated responseReduces response time for known attack patterns
OptionalPKI automationReduces certificate management overhead

1.2 Components and Functions

Each component in the hardening architecture has specific inputs, outputs, key performance indicators (KPIs), and mismatch risks. Understanding these characteristics is essential for procurement, integration, and ongoing operations. The component card diagram below provides a visual overview of all nine primary components and their operational parameters.

System Component Card Grid — Inputs, Outputs, KPIs, and Mismatch Risks
Figure 1.2: System Component Card Grid — Inputs, Outputs, KPIs, and Mismatch Risks for All Nine Primary Components

The component checklist below provides detailed specifications for each element of the hardening architecture. Each entry includes the component's primary inputs and outputs, the KPIs used to verify correct operation, and the mismatch risks that arise when the component is misconfigured, undersized, or absent.

ComponentInputsOutputsKPIsMismatch Risks
Core Switch Routing adjacencies, management sessions, ACL/CoPP policies Forwarding, logs, telemetry CPU <40% steady; control-plane drop counters stable; forwarding loss <design Insufficient TCAM for ACLs; weak CoPP; limited VRF features
OOB Mgmt Switch Mgmt ports, jump host access Isolated mgmt connectivity Mgmt network availability; port security Unmanaged switch enables MITM; no port isolation
Mgmt Firewall/ACL Gateway Admin VPN/jump host traffic Enforced allowlist to mgmt IPs Rule hit-rate; deny log volume Overly broad permits expose management plane to production traffic
Jump Host / Bastion Operator auth, MFA tokens Controlled sessions, session recording Session logging completeness; patch level; MFA enforcement rate Direct admin access bypasses audit trail; compromised jump host = full access
AAA (TACACS+/RADIUS) Credentials, role assignments AuthN/AuthZ decisions, command accounting Availability >99.9%; response latency <50ms; fail-open/close policy Misconfigured fallback locks out operations or weakens controls; no command audit
Syslog/SIEM Device logs, SNMP traps, alerts Correlation, retention, compliance reports Log completeness; timestamp accuracy; ingest latency Missing logs break audit and compliance; delayed detection of attacks
NTP System Upstream time sources (GPS/stratum-1) Consistent time to all devices Drift <100ms typical; stratum policy compliance Time skew invalidates log correlation; certificate validation failures
Backup/Config Repo Device configurations, OS images Config restores, version comparison, backup status Backup success rate; RTO for restore; storage utilization No rollback capability; config drift undetected; compliance evidence missing
Monitoring Platform Performance metrics (SNMP/telemetry), health status Real-time dashboards, performance alerts, trend analysis Polling frequency; alert fidelity; dashboard refresh rate Monitoring blind spots; alert fatigue from false positives; slow MTTR

Integration Dependencies

The hardening system components are not independent—they form a dependency chain where failure of one component can cascade to others. The most critical dependency chains are: AAA failure can cause management lockout if fallback policy is misconfigured; NTP failure can cause log correlation failures and certificate validation errors; OOB network failure can leave the core switch unreachable for emergency access; and config backup failure removes the ability to recover from configuration corruption or unauthorized changes.

Design Principle: Every core component must have a defined failure mode and fallback behavior. AAA must have a documented break-glass local account. OOB must have a console server for emergency access. Config backup must have an automated verification job. NTP must have at least two independent sources.

Deployment Checklist

Before deploying the hardening baseline, verify that all core components are present and operational. The following checklist covers the minimum requirements for a compliant deployment.

  1. OOB management network is physically or logically separate from production VLANs and has its own IP addressing plan.
  2. Management firewall rules are configured with deny-all default and explicit allow for jump host source IPs only.
  3. Jump host cluster (minimum two for HA) is deployed with MFA and session recording enabled.
  4. AAA servers (minimum two for HA) are reachable from the management VRF with shared secret rotation policy defined.
  5. Syslog/SIEM is reachable from the core switch management VRF with retention policy configured per compliance requirements.
  6. NTP servers (minimum two) are reachable from the management VRF with peer ACLs restricting to internal sources only.
  7. Config backup repository is reachable with automated backup job and restore test procedure documented.
  8. Monitoring platform has SNMPv3 or streaming telemetry configured with alert thresholds for CPU, memory, interface errors, and CoPP drops.