Zero-Trust Architecture in Practice: A Canadian Enterprise Case Study
How a mid-market Canadian financial services firm eliminated lateral movement risk, achieved SOC 2 Type II, and reduced mean time to detect from 72 hours to 4 minutes.
The perimeter is gone — most organisations have not caught up
The traditional network security model — hard shell, soft interior — has been obsolete since the widespread adoption of cloud infrastructure and remote work. Yet the majority of Canadian mid-market enterprises still operate with implicit trust inside the network perimeter: once a device or user is on the VPN, they can reach most internal resources with minimal friction. This model fails catastrophically when an attacker achieves initial access through a phishing email, a compromised contractor credential, or a vulnerable internet-facing application.
Zero-trust architecture (ZTA) replaces implicit trust with continuous verification: every access request — regardless of network location — is authenticated, authorised, and logged. This case study documents how a 1,200-employee Canadian financial services firm implemented ZTA across its Linux-heavy infrastructure over 14 months, the technical decisions made along the way, and the measurable security outcomes achieved.
The Starting Point: A Flat Network with Broad Implicit Trust
Before the engagement, the client's network architecture was typical of organisations that had grown through acquisition: a flat Layer 2 network with minimal segmentation, a site-to-site VPN connecting three offices and two data centres, and Active Directory as the sole identity provider. Linux servers (running RHEL 8 and Ubuntu 22.04) accounted for 68% of the server estate and were managed through a combination of SSH key pairs and a legacy Puppet configuration management system.
- 1,847 active SSH key pairs across the Linux estate — fewer than 30% had been rotated in the past 12 months
- No network micro-segmentation: a compromised workstation could reach any server on TCP/22 and TCP/443
- Mean time to detect (MTTD) for lateral movement: 72 hours (based on tabletop exercise results)
- No privileged access workstation (PAW) controls — administrators used their standard workstations for privileged operations
- Audit logging was inconsistent: 40% of Linux servers shipped logs to a central SIEM; the remainder logged locally only
The Zero-Trust Architecture Design
The ZTA design was built on four pillars: identity-centric access, micro-segmentation, privileged access management, and continuous monitoring. The implementation was phased to avoid operational disruption — the client could not afford a 'big bang' cutover given the criticality of their trading and settlement systems.
- Identity provider: Okta Workforce Identity, federated with Active Directory via LDAP sync — single source of truth for all human and service identities
- Network micro-segmentation: Illumio Core deployed on all Linux and Windows servers, enforcing allow-list policies based on application dependency maps generated during a 30-day discovery phase
- Privileged access management: CyberArk Privileged Access Manager replaced SSH key pairs — all privileged Linux access routed through CyberArk with session recording and just-in-time provisioning
- Device trust: Jamf Pro (macOS) and Microsoft Intune (Windows) enforced device compliance posture as a condition of network access; Linux workstations enrolled in a custom compliance check via Osquery
- SIEM: Splunk Enterprise Security with UEBA (User and Entity Behaviour Analytics) — all Linux audit logs, CyberArk session data, and Illumio flow logs centralised and correlated
Linux-Specific Implementation Challenges
Linux environments present unique challenges for zero-trust implementation that Windows-centric security tools often handle poorly. Three issues required custom engineering work during this engagement.
- Service account proliferation: 340 service accounts were discovered during the identity audit — many were shared across applications and had not been reviewed in years. Each was mapped to a specific application, rotated into CyberArk, and granted only the minimum permissions required
- SSH key remediation: all 1,847 SSH key pairs were revoked over a 6-week period, replaced with CyberArk-managed ephemeral certificates with 8-hour TTLs. The transition required coordination with 23 application teams to update deployment pipelines
- Illumio policy tuning: the initial micro-segmentation policy generated 4,200 policy violations in the first week — mostly legitimate application traffic that had not been documented. A 45-day policy refinement phase reduced violations to 12 (all genuine anomalies) before enforcement mode was activated
Incident Response Integration
Zero-trust architecture generates significantly more telemetry than traditional perimeter-based security — which is only valuable if the SOC can act on it quickly. The client's existing incident response process was manual and undocumented, with a mean time to respond (MTTR) of 4.5 hours for high-severity alerts. The ZTA implementation included a parallel IR process redesign.
- Splunk SOAR (Security Orchestration, Automation, and Response) automated the first 15 minutes of response for the 12 highest-frequency alert types — account lockout, impossible travel, and lateral movement indicators
- Playbooks were documented for 28 alert types, with clear escalation paths and decision trees for L1 analysts
- Tabletop exercises were conducted monthly for the first 6 months post-implementation to validate playbook effectiveness
- A 24/7 managed detection and response (MDR) retainer was established with a Canadian MSSP for after-hours coverage
Measurable Outcomes After 12 Months
Twelve months after the ZTA implementation was complete, the client underwent a SOC 2 Type II audit covering the full 12-month period. The results validated the investment.
- SOC 2 Type II certification achieved with zero exceptions in the access control and monitoring trust service criteria
- Mean time to detect (MTTD) reduced from 72 hours to 4 minutes for lateral movement indicators
- Mean time to respond (MTTR) reduced from 4.5 hours to 22 minutes for high-severity alerts
- SSH key pairs eliminated: 100% of privileged Linux access now routed through CyberArk with full session recording
- Micro-segmentation coverage: 98.7% of server-to-server traffic now governed by explicit allow-list policies
- PIPEDA compliance posture strengthened: the client's legal team confirmed the ZTA architecture satisfies the technical safeguard requirements for the personal financial data processed on the Linux estate
Key Outcomes at 12 Months
Zero-trust is an architecture, not a product
The most important lesson from this engagement is that zero-trust is an architectural approach, not a product you can purchase and deploy. The technology — Okta, Illumio, CyberArk, Splunk — is the implementation layer. The hard work is the identity audit, the application dependency mapping, the policy tuning, and the process redesign. Organisations that treat ZTA as a product procurement exercise consistently underestimate the effort and overestimate the speed of deployment.
Alhassan Consultancy's enterprise security practice helps Canadian organisations design and implement zero-trust architectures that satisfy regulatory requirements and withstand real-world adversaries. Book a confidential security assessment to benchmark your current posture.