Network security best practices for Canadian SMBs start with three non-negotiables: a properly configured next-generation firewall with default-deny outbound rules, network segmentation that isolates staff, guests, IoT and servers into separate VLANs, and enforced MFA on every identity. Layer in endpoint detection, secure Wi-Fi, and log monitoring to close the remaining gaps — and you eliminate the attack paths behind more than 90% of incidents the Canadian Centre for Cyber Security documents each year.
Why Canadian SMBs Are the Prime Network Target
The Canadian Centre for Cyber Security's 2024 National Cyber Threat Assessment is blunt: small and medium businesses are the most commonly targeted segment of the Canadian economy. They hold valuable data — customer PII, financial records, supplier contracts — but typically lack the dedicated security staff of a large enterprise. That asymmetry makes them attractive.
CIRA's 2023 Cybersecurity Survey found that 30% of Canadian organizations experienced a cyber incident in the prior 12 months, with SMBs disproportionately represented. The average cost of a data breach in Canada reached CA$6.32 million in 2023 (IBM Cost of a Data Breach Report, Canada), with small organizations bearing a proportionally larger share of that cost relative to revenue.
The good news is that the attack paths are well-documented and largely preventable. The Canadian Centre for Cyber Security (cyber.gc.ca) consistently identifies the same root causes year after year: unpatched software, missing MFA, poorly configured firewalls, flat networks that allow lateral movement, and unmonitored endpoints. None of those gaps require a six-figure security budget to close. They require deliberate configuration and consistent operational discipline.
A manufacturing company in Hamilton, a law firm in Calgary, a dental clinic in Laval — they all share the same fundamental network exposures, and the same remediation playbook applies. What follows is that playbook, structured from the perimeter inward.
The Layered-Defence Model (Defence-in-Depth)
Defence-in-depth means stacking independent security controls so that the failure of any single layer does not result in a complete breach. No single technology is impenetrable — the goal is to make attackers expend enough effort that they move to easier targets, or to detect them before they accomplish their objective.
For a Canadian SMB, the practical layers are:
- Perimeter layer: Next-generation firewall (NGFW) with IPS, DNS filtering, and application-aware rules.
- Network layer: VLAN segmentation, 802.1X port authentication, private VLANs for sensitive segments.
- Access layer: MFA on every identity, role-based access control (RBAC), least-privilege accounts.
- Endpoint layer: Endpoint detection and response (EDR), disk encryption (BitLocker/FileVault), patch management.
- Wireless layer: WPA3-Enterprise (or WPA2-Enterprise minimum), SSID isolation, rogue AP detection.
- Remote access layer: ZTNA or hardened VPN with MFA, device posture checks.
- Monitoring layer: Centralized logging, SIEM correlation, or managed detection and response (MDR).
- Recovery layer: Immutable offsite backups, tested restore procedures — covered in the backup and disaster recovery guide.
Each subsequent section of this guide covers one layer in detail. The order is deliberate: start at the perimeter and work inward. Reconfiguring a firewall costs nothing; recovering from ransomware that moved laterally across a flat network costs on average CA$178,000 in direct costs for Canadian SMBs (Insurance Bureau of Canada, 2023).
Firewall Configuration: The Foundation Layer
A firewall that ships with factory defaults is not a security control — it is a false sense of security. Most business-grade firewalls (Fortinet FortiGate, Sophos XGS, Cisco Meraki MX, Ubiquiti UniFi Gateway) default to allow-all outbound and block-inbound-unsolicited, which stops opportunistic inbound scans but does nothing to prevent an infected workstation from phoning home to a command-and-control server or exfiltrating data.
The following configuration steps apply to virtually every business-class NGFW:
- Enable the IPS/IDS subscription. Most NGFWs include an intrusion prevention signature feed as a licensed add-on. Activate it. Inspect both inbound and outbound traffic. Performance impact is negligible on modern hardware.
- Switch to default-deny outbound. Create explicit permit rules only for traffic your business needs: port 443 (HTTPS), port 80 (HTTP, via HTTPS redirect), port 25/587/993 (email), DNS on port 53 to your DNS provider only. Everything else — deny and log.
- Enable application-layer inspection (NGFW policy). Identify and block high-risk application categories: Tor, anonymizers, peer-to-peer file sharing, cryptocurrency mining. Block country groups you have no business reason to communicate with — Canadian SMBs rarely need direct connections to high-risk country ASNs.
- Enable DNS filtering. Route all DNS queries through a protective DNS resolver. CIRA Canadian Shield (shield.cira.ca) is a free, Canadian-hosted option that blocks known malicious domains without sending your query data offshore. Cisco Umbrella and Cloudflare Gateway are alternatives with more granular policy controls.
- Lock down management access. The firewall admin interface must only be reachable from a dedicated management VLAN, not from the general staff network. Disable remote (WAN-side) admin access entirely unless you have a specific monitored use case — and even then, restrict it to specific source IPs and require MFA.
- Enable logging to a centralized destination. Firewall logs are only useful if they go somewhere. Configure syslog export to your log management platform (see the monitoring section). Enable logging for all deny events at minimum; log permit events for internet-facing rules.
- Schedule firmware updates quarterly. Firewall vendors issue security advisories regularly — Fortinet, Palo Alto, and SonicWall have all had high-severity CVEs exploited in the wild in the past two years. A firmware update cycle aligned with your quarterly rule review keeps the underlying platform current.
For offices with more than 20 staff, consider a hardware-based NGFW rather than a software firewall on a general-purpose server — purpose-built silicon handles deep packet inspection at full throughput without degrading user experience. Entry-level hardware NGFWs suitable for a 20–50 person office (Fortinet FortiGate 60F, Sophos XGS 87, Cisco Meraki MX67) retail for CA$800–$1,800 with a one-year security subscription.
Network Segmentation and VLANs
A flat network — one where every device shares the same Layer 2 broadcast domain — means that a ransomware payload on a receptionist's PC can scan and reach your accounting server, your NAS, and your SCADA controller (if applicable) without crossing any enforced security boundary. VLANs eliminate this lateral movement path by dividing the network into isolated segments that must traverse the firewall to communicate.
The minimum segment structure for any Canadian SMB:
- Staff VLAN (VLAN 10): All employee workstations and laptops. Outbound internet access; deny inbound from all other VLANs except the management VLAN.
- Server/NAS VLAN (VLAN 20): File servers, domain controllers, NAS, backup appliances. Only staff and management VLANs can initiate connections, on specific permitted ports only.
- Guest/IoT VLAN (VLAN 30): Guest Wi-Fi, smart TVs, printers, IP cameras, building automation systems. Internet access only — no access to any other VLAN. This is the segment where most IoT compromise happens, and it must be completely firewalled from everything else.
- VoIP VLAN (VLAN 40, if applicable): IP phones and SIP trunks. Separate QoS policy; restrict to the SIP provider's IPs only.
- Management VLAN (VLAN 99): Network switches, access points, firewalls, out-of-band management interfaces. Reachable only from the IT administrator's workstation. No end-user devices on this VLAN, ever.
Implementation requires a managed switch (unmanaged switches cannot support VLANs). Entry-level 24-port managed switches from Cisco Catalyst, Netgear MS, or Ubiquiti UniFi cost CA$250–$700 and are sufficient for an office of 20–30 staff. Configure inter-VLAN routing on the firewall — not on the switch — so that all cross-segment traffic passes through your security policy.
802.1X port authentication (using your Active Directory or a RADIUS server like Windows NPS) ensures that only authenticated devices can join a port on the correct VLAN. Without 802.1X, a visitor who plugs a cable into a wall jack joins your staff VLAN. With it, an unauthorized device gets dropped to the guest VLAN automatically. This control is often skipped by small offices due to complexity — but it is available natively in Windows Server and most managed switches, and a competent IT provider can deploy it in a day.
VPN vs. ZTNA: Choosing the Right Remote Access Model
Remote access security is where most Canadian SMBs made their biggest security debt during the 2020–2022 work-from-home expansion. A site-to-site VPN or a legacy SSL VPN grants connecting devices a seat on your internal network — which means a compromised remote laptop is now inside your perimeter. Zero Trust Network Access (ZTNA) replaces that model with per-application, identity-verified, device-posture-checked access.
| Criteria | Traditional VPN | ZTNA |
|---|---|---|
| Access model | Full network tunnel | Per-application access |
| Lateral movement risk | High — connected device joins internal network | Low — access limited to authorized apps only |
| Device posture check | Optional, inconsistent | Built-in, enforced continuously |
| Best for | On-premises apps, OT/SCADA access | Cloud-first SMBs, M365/SaaS environments |
| Typical CA$ cost/user/mo | $5–$15 (software client); $0 if using firewall VPN | $12–$35 (Cloudflare Access, Zscaler Private Access, Entra ID P1) |
| Setup complexity | Low–medium | Medium (requires identity integration) |
| MFA enforcement | Optional (must be explicitly configured) | Mandatory by design |
For Canadian SMBs running Microsoft 365, the simplest ZTNA path is Entra ID (formerly Azure AD) Conditional Access with device compliance policies — already included in Microsoft 365 Business Premium (CA$28.10/user/month in 2026). It grants access to M365 apps only from compliant, Intune-enrolled devices, requires MFA on every sign-in, and blocks access from high-risk sign-in locations. That covers the majority of the remote-access surface without any additional tooling.
If you have on-premises file servers or legacy applications that cannot authenticate through a cloud identity provider, retain an IPSec or SSL VPN for those specific resources — but restrict it to specific applications and subnets, not full network access. The remote work security guide covers hybrid VPN/ZTNA architectures in depth.
Endpoint Protection and EDR
The network perimeter stops what it can see. The endpoint is where what gets through must be contained. Signature-based antivirus alone — the kind that shipped with Windows 7 and still ships on most consumer-grade devices — detects known malware by file hash. It misses fileless attacks, living-off-the-land techniques, and zero-day exploits that have not yet generated a signature. Endpoint detection and response (EDR) replaces that model with behavioural analysis: it watches what processes do, not just what they look like.
Practical requirements for a Canadian SMB endpoint strategy:
- Deploy EDR, not just antivirus. Microsoft Defender for Business is included in M365 Business Premium and provides EDR capability adequate for most SMBs. Third-party options (CrowdStrike Falcon Go, SentinelOne Singularity, Malwarebytes ThreatDown) cost CA$8–$20/endpoint/month and offer more granular telemetry and threat-hunting capability.
- Enable full-disk encryption on every device. BitLocker (Windows) and FileVault (macOS) are built-in, free, and encrypt the drive if a laptop is stolen. A stolen unencrypted laptop containing client data is a PIPEDA breach reportable to the Office of the Privacy Commissioner. An encrypted stolen laptop is not — as long as keys were managed correctly.
- Enforce automated patching. Unpatched OS and application vulnerabilities are the most common network entry point after phishing. Windows Update for Business (free) or a third-party patch management tool (NinjaOne, Automox, Kaseya VSA) can push patches and report compliance from a central console. Target: all critical patches within 72 hours of release.
- Apply application control / allowlisting. Block execution of software not on an approved list. This is a tier-2 control — complex to implement without causing user friction — but it is the single most effective control against ransomware that arrives as an email attachment. Microsoft Defender Application Control (MDAC) and AppLocker are available natively in Windows 10/11 Pro and Enterprise.
- Manage mobile devices (MDM/MAM). Any employee smartphone or tablet that receives company email or accesses cloud apps must be enrolled in a mobile device management platform (Intune, Jamf) or at minimum have mobile application management (MAM) policies applied — so that a "wipe corporate data" command can be executed remotely if the device is lost or the employee leaves.
Securing Business Wi-Fi
Business Wi-Fi is a significant attack surface that is frequently misconfigured. Consumer-grade routers in a commercial environment, a single SSID for staff and guests, default admin passwords, WPA2-Personal with a shared passphrase posted in the reception area — these are the configurations seen in the majority of Canadian SMB assessments. Each one is a readily exploitable entry point.
Minimum Wi-Fi security configuration for a Canadian SMB:
- Use WPA3 or WPA2-Enterprise. WPA3 (802.11ax/Wi-Fi 6 access points support it natively) eliminates the offline dictionary attack vulnerability in WPA2-Personal. WPA2-Enterprise with 802.1X is the alternative for older AP hardware — each user authenticates with their Active Directory credentials, so there is no shared passphrase to leak.
- Separate guest and staff SSIDs at Layer 2. Guest SSID maps to the Guest/IoT VLAN (VLAN 30) with internet-only access. Staff SSID maps to the Staff VLAN (VLAN 10). These must be on different VLANs with a firewall enforcing the boundary — an AP that supports two SSIDs but maps them to the same VLAN provides zero isolation.
- Disable legacy protocols. Disable WPS (Wi-Fi Protected Setup) — it has a brute-force vulnerability (Pixie Dust) that can recover the network passphrase in minutes. Disable 802.11b/g support if all devices are 802.11n or newer — legacy protocol support increases attack surface with no upside.
- Enable rogue AP detection. A rogue access point — a device brought in by an employee or planted by a threat actor — is broadcasting on your channel and potentially intercepting traffic. Business-grade Wi-Fi platforms (Cisco Meraki, Ubiquiti UniFi, Aruba Instant) include rogue AP detection and alerting as a standard feature.
- Change default SSID and admin credentials immediately. An SSID that includes the manufacturer name (Netgear-1234, ASUS_ROUTER) or office address reveals information. Default admin credentials for most consumer and semi-commercial AP brands are published online and are trivially exploited.
- Enable client isolation on the guest SSID. Client isolation prevents devices on the guest network from communicating with each other — so a guest laptop running malware cannot attack other guest devices. This is a one-checkbox setting in virtually all managed Wi-Fi platforms.
For offices with 10+ concurrent wireless users, invest in business-grade access points with centralized management. Ubiquiti UniFi U6 Pro access points (CA$220 each) or Meraki MR46 (CA$900+ with license) both support the configuration requirements above and can be managed from a cloud dashboard.
Identity and Access Management (MFA + Least Privilege)
Identity is the new perimeter. In an environment where most applications are cloud-hosted and accessible from any device on the internet, a stolen credential bypasses every network control you have built. The Canadian Centre for Cyber Security identifies credential compromise as the leading initial access vector in Canadian incidents. The two controls that address this most directly are multi-factor authentication and the principle of least privilege.
Multi-factor authentication (MFA): Enable phishing-resistant MFA — hardware security keys (YubiKey, Google Titan) or authenticator app TOTP (Microsoft Authenticator, Google Authenticator) — on every account that accesses corporate data. SMS-based MFA is better than nothing but is vulnerable to SIM-swapping. Prioritize: email and Microsoft 365 first, then cloud file storage, then banking and payroll, then every other login. See the cybersecurity guide for MFA deployment details.
Least privilege: Every user account should have only the permissions required for that user's job function — nothing more. Standard employees should not be local administrators on their PCs. Nobody should be logging in daily with a domain administrator account. Privileged accounts (domain admin, global admin, root) should be separate accounts used only for administrative tasks, with MFA enforced, and with all privileged sessions logged. Review permissions quarterly and revoke access for departed employees within the same business day they leave.
Shared accounts are a dead-end. Any account shared by multiple users — a generic "office@company.ca" that three people use, a single admin password on the firewall known to all IT staff — cannot be audited, cannot be tied to an individual in a breach investigation, and cannot be revoked cleanly when one person leaves. Eliminate shared credentials.
Network Monitoring and Log Management
A breach that goes undetected for 200 days — the IBM global average dwell time — causes exponentially more damage than one detected in 48 hours. For Canadian SMBs under PIPEDA, the clock on your mandatory breach notification to the Office of the Privacy Commissioner (priv.gc.ca) begins when you become aware of the breach — not when it started. Earlier detection means lower notification penalties, lower remediation cost, and a more defensible posture with regulators.
Log management and monitoring for a Canadian SMB:
- Centralize logs from all critical sources. Firewall deny/permit logs, Windows Event Logs (especially 4624/4625 login events, 4720 account creation, 7045 service installation), DNS query logs, and cloud audit logs (Microsoft 365 Unified Audit Log, Azure AD sign-in logs). Even a free ELK stack (Elasticsearch, Logstash, Kibana) on an internal VM provides centralized search; commercial SIEMs (Microsoft Sentinel, Elastic Security, Huntress SIEM) add correlation rules.
- Set retention to 12 months minimum. PIPEDA incident investigations frequently require log evidence from months before the discovered event. Logs retained for only 30 days (a common default) cannot support a forensic investigation. Twelve months of compressed firewall and event logs for a 30-person office typically requires less than 50 GB of storage.
- Create baseline alerts. Even without a full SIEM, basic alerting on: any new local admin account created, any RDP connection from an external IP, any firewall rule change, any DNS query to a new TLD outside your normal set. Most NGFWs support email alerts; Windows Event Viewer supports XML filter subscriptions.
- Consider managed detection and response (MDR). For organizations without dedicated security staff, an MDR service monitors your telemetry 24/7 and provides human triage on alerts. Canadian-managed MDR options through IT service providers typically run CA$15–$40 per endpoint per month — a cost that is almost always less than the incident response bill from a missed alert.
Network Security Tool Costs in Canada (2026, CA$)
The following table provides vendor-neutral indicative pricing ranges for the controls described in this guide. Actual quotes depend on scope, contract length, and number of users — treat these as order-of-magnitude benchmarks for budget planning.
| Control | Entry option (free/bundled) | Paid range (CA$/mo) | Notes |
|---|---|---|---|
| Next-gen firewall (hardware) | pfSense CE (open source) | $80–$200/mo (hardware amortized + subscription) | One-time hardware CA$800–$1,800 + annual security subscription |
| DNS filtering | CIRA Canadian Shield (free) | $3–$8/user/mo (Cisco Umbrella, Cloudflare Gateway) | CIRA Shield is Canadian-hosted, free for all Canadians |
| EDR / endpoint protection | Defender for Business (M365 BP) | $8–$22/endpoint/mo (standalone) | M365 Business Premium (CA$28.10/user/mo) includes Defender for Business |
| ZTNA / conditional access | Cloudflare Access (free ≤50 users) | $12–$35/user/mo (Zscaler, Entra ID P1) | Entra ID P1 included in M365 Business Premium |
| Managed detection & response (MDR) | Huntress (includes SIEM + MDR) | $15–$40/endpoint/mo | Often bundled by Canadian MSPs; displaces need for internal SOC analyst |
| Business Wi-Fi (hardware) | Ubiquiti UniFi U6 Lite (CA$150/AP) | CA$220–$900/AP (hardware, one-time) | UniFi cloud controller is free; Meraki requires per-device license CA$30–$90/AP/year |
| Full stack (MSP-managed) | — | $40–$100/user/mo | All controls above, managed. Excludes hardware one-time costs. |
The full managed stack — firewall, DNS filtering, EDR, ZTNA, MDR — runs CA$40–$100 per user per month through a Canadian managed IT services provider. For a 20-person office, that is CA$800–$2,000 per month. Against a median Canadian SMB ransomware recovery cost of CA$178,000, the ROI is not subtle.
Seven Common Network Security Configuration Mistakes
These are the misconfigurations found consistently in Canadian SMB security assessments. Each one is avoidable and each one is exploited regularly.
- Firewall with factory defaults, never touched after install. The device is physically present; the policy is not. Default configurations allow all outbound traffic, which means an infected workstation can freely communicate with a command-and-control server. Every NGFW deployed must be configured by someone who reads the documentation.
- Single flat network for everything. Staff, guests, IoT devices, servers, printers all on the same subnet. A flat network amplifies every infection that gets through the perimeter. Segmentation is not a premium feature — any managed switch supports VLANs, and most NGFWs support inter-VLAN firewall policy at no additional cost.
- VPN without MFA. A VPN with only a username and password is nearly worthless — credentials are phished or brute-forced routinely. Every remote access connection must require a second factor, every time.
- Guest Wi-Fi on the same VLAN as staff devices. A guest SSID that maps to the same network as your workstations provides zero isolation. If a guest laptop is infected, it has full Layer 2 access to every device on that network.
- Administrator accounts used for daily tasks. Domain admins checking email is a textbook attack path. A phishing email opened in a domain admin session means the attacker immediately has the highest privilege on your network. Standard accounts for daily use; privileged accounts used only for administrative sessions.
- EDR alerts not monitored. Deploying an EDR tool and not monitoring its alerts is worse than not deploying it — it creates a false sense of coverage. If your team cannot investigate EDR alerts, engage a managed detection service that will.
- No offboarding procedure. A departed employee's account left active for even 24 hours is a live credential that may be used maliciously — either by the former employee or by a threat actor who purchased their credentials on the dark web. Revoke access on the day of departure, every time, without exception.
Canadian Regulatory Context: PIPEDA, Law 25, and Sector Rules
Canadian network security is not just an operational matter — it intersects directly with legal obligations under several federal and provincial frameworks. Non-compliance is no longer theoretical: the Office of the Privacy Commissioner of Canada (priv.gc.ca) has increased enforcement activities, and Quebec's Commission d'accès à l'information has issued six-figure fines under Law 25 since its third phase came into force in September 2023.
PIPEDA (federal): Requires organizations to protect personal information with safeguards appropriate to the sensitivity of the information. A breach of security safeguards involving a real risk of significant harm must be reported to the OPC and affected individuals. The regulation does not specify which controls are required, but the OPC's published findings make clear that missing MFA, unencrypted devices, and flat networks have been cited in breach investigations as evidence of inadequate safeguards.
Quebec Law 25 (provincial, strongest in Canada): Requires organizations operating in Quebec to appoint a privacy officer, conduct privacy impact assessments (PIAs) for high-risk projects, obtain meaningful consent for data collection, and notify the CAI within 72 hours of a breach involving personal information. The most directly relevant network security requirement is the explicit obligation to implement security measures proportionate to the sensitivity of the data — including technology controls. The Law 25 compliance guide covers this framework comprehensively.
PHIPA (Ontario, health data): Ontario health information custodians must implement administrative, technical, and physical safeguards for personal health information. The Information and Privacy Commissioner of Ontario has explicitly cited failure to segment networks and failure to implement access controls in healthcare breach findings.
PCI DSS (payment card data): Any business that accepts payment cards is subject to PCI DSS. Version 4.0 (mandatory from March 2025) requires network segmentation between the cardholder data environment and other networks, firewall rules restricting traffic into and out of the CDE, and multi-factor authentication for all CDE access. Failure to comply is a contractual issue with your payment processor and can result in card acceptance being revoked.
CCCS guidance: The Canadian Centre for Cyber Security publishes actionable guidance at cyber.gc.ca — including the ITSP.10.171 Small and Medium Organizations guidance and the Baseline Cyber Security Controls for Small and Medium Organizations. These are not legally binding but are the de facto standard for demonstrating due diligence in Canadian breach investigations.
Case Study: Toronto Professional Services Firm, 28 Staff
This anonymized case study reflects a composite of real engagements. The firm — a mid-size accounting practice in downtown Toronto with 28 staff, a mix of Windows 10 and macOS devices, and a Microsoft 365 subscription — experienced a ransomware incident in late 2024 that encrypted their on-premises file server and three workstations.
What happened: A bookkeeper received a phishing email that impersonated CRA's business account portal. The link led to a credential harvesting page, and her M365 credentials were captured. Because the firm had no MFA and no Conditional Access policy, the attacker authenticated directly to SharePoint and downloaded three months of client financial records. The bookkeeper's account was also used to send further phishing emails to the firm's clients. Two days later, the attacker deployed ransomware through a remote desktop session (RDP was open to the internet on the file server, on the default port 3389, with a guessable service account password).
Root causes identified: No MFA; RDP exposed to the internet; flat internal network with no VLAN separation; no EDR on the file server; no log monitoring — the RDP brute-force activity ran for 11 days before the ransomware detonated, completely undetected.
Total incident cost: Approximately CA$210,000, including CA$65,000 in ransomware recovery and forensics, CA$80,000 in lost billable hours during the 19-day recovery, CA$35,000 in client notification and credit monitoring, and CA$30,000 in regulatory response and legal fees. The firm had a cyber insurance policy but the claim was disputed due to unimplemented basic controls the policy required.
Post-incident controls deployed: M365 Business Premium with Conditional Access and MFA (CA$28.10/user/month); network segmentation with VLANs and a Fortinet FortiGate 60F (CA$1,400 hardware + CA$480/year subscription); EDR via Defender for Business; RDP disabled externally (replaced with Entra ID app proxy for remote access); Huntress MDR (CA$22/endpoint/month). Total ongoing spend: approximately CA$62/user/month. Annual cost increase: approximately CA$14,000. Avoided incident cost: CA$210,000.
The firm engaged the Quebec and Ontario on-site delivery team at IT Cares for hands-on firewall and endpoint hardening — the configuration work took two days on-site and one remote follow-up session. The managed services engagement has prevented two subsequent credential-stuffing attempts that were blocked and alerted on within minutes of the attack.
The 30-Point Network Security Checklist
Use this checklist to assess your current posture. Any unchecked item is an active risk. Prioritize the first ten before moving to the rest.
Firewall & Perimeter
- Next-generation firewall deployed (not consumer router)
- Default-deny outbound policy configured
- IPS/IDS subscription enabled and active
- DNS filtering enabled (CIRA Shield or equivalent)
- Application control policy blocking high-risk categories
- Management interface accessible from management VLAN only
- Firewall firmware updated within last 90 days
- Firewall rules reviewed within last quarter; stale rules removed
Network Segmentation
- Managed switches deployed (not unmanaged)
- VLANs configured: Staff, Servers, Guest/IoT, Management, VoIP
- Inter-VLAN routing enforced by firewall, not switch
- Guest devices have internet-only access, no internal routing
- IoT devices (cameras, HVAC, printers) on isolated VLAN
- 802.1X port authentication enabled on wired ports (or compensating control documented)
Identity & Access
- MFA enabled on every cloud account (email, M365, banking, payroll)
- MFA enabled on VPN or remote access solution
- No shared credentials (every user has a unique account)
- Daily users not running as local administrator
- Domain admin / global admin accounts used only for administrative tasks
- Departed employee accounts revoked within 24 hours of departure
- Access review conducted within last 6 months
Endpoint & Wi-Fi
- EDR deployed on all workstations, laptops and servers
- Full-disk encryption enabled on all laptops (BitLocker/FileVault)
- Critical patches applied within 72 hours of release
- Mobile devices enrolled in MDM or MAM policy
- Business-grade Wi-Fi AP (WPA3 or WPA2-Enterprise)
- Separate SSID for guests/IoT, mapped to isolated VLAN
- WPS disabled on all access points
Monitoring & Recovery
- Logs centralized with 12-month minimum retention
- Alerts configured for critical events (new admin, RDP, rule changes)
- EDR alerts assigned to a person who investigates them
- Offsite, immutable backups tested for restore in last 90 days (see backup guide)
Frequently Asked Questions
What are the most important network security best practices for a Canadian SMB?
Start with three controls: enforce multi-factor authentication on every account, deploy a next-generation firewall with default-deny outbound rules, and segment your network with VLANs so that a breach in one area cannot spread. Together these stop the vast majority of incidents documented by the Canadian Centre for Cyber Security each year.
How much does network security cost for a small business in Canada?
A complete managed stack — next-gen firewall, DNS filtering, EDR, ZTNA, and log monitoring — runs roughly CA$40–$100 per user per month when managed by a Canadian IT provider. Hardware (firewall, switches, access points) is typically CA$2,000–$6,000 one-time for an office of 20–50 staff. The median cost of a ransomware incident in Canada is CA$178,000 — the math is straightforward.
Does my 10-person office need VLANs?
Yes. Even a 10-person office benefits from three segments: a staff VLAN, a guest/IoT VLAN, and a server VLAN. Without segmentation, a compromised smart TV or guest laptop sits on the same Layer 2 network as your accounting server and file shares. The configuration takes under an hour with a managed switch and most business-grade routers — it is not a complex project.
Should I replace my VPN with ZTNA?
If your workforce is primarily cloud-based (Microsoft 365, Salesforce, QuickBooks Online) with 10–50 staff, ZTNA is the better long-term choice — it grants per-application access rather than full network tunnel access, eliminating lateral movement risk. For significant on-premises workloads, retain a VPN for those segments and layer ZTNA for everything else. Microsoft 365 Business Premium includes the Entra ID Conditional Access needed for a ZTNA baseline at no additional cost.
What does PIPEDA require for network security?
PIPEDA requires "appropriate safeguards" proportionate to the sensitivity of data — there is no prescriptive control list. However, the Office of the Privacy Commissioner of Canada (priv.gc.ca) consistently cites missing MFA, unencrypted devices, and flat networks in breach investigations as evidence of inadequate safeguards. Quebec's Law 25 adds a 72-hour mandatory breach notification requirement and an explicit obligation to appoint a privacy officer. A breach caused by a missing basic control is nearly indefensible before the OPC.
How often should firewall rules be reviewed?
At minimum, quarterly. Firewall rule sets accumulate stale permit rules — a former vendor's access, a past project, a one-time exception never removed. Each stale rule is a residual attack surface. Audit any rule that permits inbound traffic from the internet every time a new rule is added, and review the full rule set on a quarterly calendar cycle.
What is the difference between IDS and IPS?
An IDS (Intrusion Detection System) monitors traffic and generates alerts — it observes and reports. An IPS (Intrusion Prevention System) sits inline and blocks malicious traffic in real time. For SMBs, an IPS integrated into a next-generation firewall (available in Fortinet FortiGate, Sophos XGS, Cisco Meraki MX) is the practical choice: it combines detection and blocking without requiring a separate appliance or a dedicated analyst to process alerts.
How do I know if my network has been compromised?
Key indicators: unexpected outbound connections to foreign IPs or unfamiliar cloud services; a spike in DNS queries to new or unusual domains; new administrator accounts appearing in Active Directory or your cloud identity provider; EDR alerts that were generated but never investigated; and staff complaints about unexplained slowness. A managed detection service (MDR) correlates these signals automatically. Without one, you're relying on manually noticing anomalies — which most SMBs miss until ransomware detonates.
Get a Free Network Security Assessment
Tell us about your current setup and we'll map the gaps. Independent guidance from TechCare Canada — hands-on delivery available across Canada.
