November 19, 2026
From CCNA Theory to Enterprise Network Engineering: Designing, Configuring and Hardening a…
Building a redundant multi-VLAN enterprise network, implementing layered security controls, and troubleshooting a misconfiguration that…

By Citadel Cybersec
14 min read
- 1 Building a redundant multi-VLAN enterprise network, implementing layered security controls, and troubleshooting a misconfiguration that connectivity tests failed to reveal.
- 2 1. Project at a glance
- 3 2. Architecture and design intent
- 4 2.1 Aligning HSRP and spanning-tree roles
- 5 2.2 Redundancy with two different paths
Building a redundant multi-VLAN enterprise network, implementing layered security controls, and troubleshooting a misconfiguration that connectivity tests failed to reveal.
After 13 years as a graphic designer, I decided to transition into IT, focusing on networking and cybersecurity.
Passing the CCNA was an important milestone, but I wanted to demonstrate something beyond exam knowledge: the ability to apply networking concepts to a practical infrastructure project, make engineering decisions, troubleshoot unexpected behavior, and verify that the resulting configuration actually works.
To put that knowledge into practice, I designed and built a redundant, collapsed-core enterprise network featuring multiple VLANs, dynamic routing, first-hop redundancy, centralized network services, and access-layer security controls.
I also encountered an interesting problem: every connectivity test passed, yet part of my NAT configuration was incorrect.
Finding the cause required looking beyond successful pings, tracing routing decisions, examining NAT translations, and identifying an artificial routing behavior in the simulated ISP that was masking the problem.
This project brings together networking, infrastructure security, troubleshooting, and the analytical approach I've developed through my blue-team training and practical cybersecurity investigations.
Here is how I designed, implemented, secured, and tested it.
1. Project at a glance
- Architecture: Redundant collapsed-core topology
- Switching: VLANs, pruned trunks, PVST+, LACP EtherChannel
- First-hop redundancy: HSRP
- Routing: OSPF Area 0, ECMP, loopback router IDs
- Network services: Centralized DHCP, DHCP relay, NAT/PAT, DNS, NTP
- Access-layer security: DHCP Snooping, Dynamic ARP Inspection, Port Security, BPDU Guard
- Management security: SSHv2, management ACLs, login protection
- Edge security: Inbound antispoofing ACL
- Environment: Cisco Packet Tracer v9.0.1.0858
- Project resources: [GitHub repository — complete configurations, topology, and verification evidence]
2. Architecture and design intent
The network follows a collapsed-core architecture, combining the core and distribution layers into a single redundant tier.
Its main components are:
- MLSW1 and MLSW2: Two multilayer switches providing inter-VLAN routing, first-hop redundancy, and the collapsed-core layer.
- SW1–SW3: Three access switches, each serving a dedicated user VLAN.
- R1: The primary edge router, responsible for centralized DHCP, NAT/PAT, and connectivity toward the ISP.
- R2: An additional interior router providing an alternative path toward the edge.
- ISP and Server0: Simulated external connectivity and network services.
The three user VLANs are VLANs 10, 20, and 30. VLAN 99 is reserved as a dedicated native VLAN, with no user endpoints assigned to it.
Two architectural decisions were particularly important.
2.1 Aligning HSRP and spanning-tree roles
I deliberately aligned the HSRP active gateway and spanning-tree root bridge roles for each VLAN.
- VLANs 10 and 30: MLSW1 is both HSRP Active and STP Root.
- VLAN 20: MLSW2 is both HSRP Active and STP Root.
For example, MLSW2 receives an increased HSRP priority for VLAN 20, alongside an STP priority of 4096.
This alignment helps ensure that the Layer 2 forwarding topology agrees with the Layer 3 gateway placement.
Without it, traffic may follow a suboptimal path through the inter-switch connection before reaching its active gateway. Aligning both roles avoids unnecessary Layer 2 traversal and makes forwarding behavior more predictable.
2.2 Redundancy with two different paths
The multilayer switches are interconnected through a Layer 3 LACP EtherChannel.
MLSW1 has a direct routed connection toward R1, whereas MLSW2 reaches R1 through R2.
This creates an interesting routing scenario: MLSW2 can reach the edge through two equal-cost OSPF paths.
Rather than implementing redundancy solely as a standby mechanism, this topology allows both paths to participate in forwarding.
That design decision becomes particularly important when investigating the NAT issue discussed in Section 6.
Each infrastructure device also has a loopback interface with a memorable IP address (1.1.1.1, 2.2.2.2, 3.3.3.3, etc.), used as its OSPF router ID.
R1's loopback additionally provides a stable internal NTP reference. Downstream devices synchronize against it, while R1 synchronizes with the external NTP server.
Using loopbacks provides stable identifiers independent of the physical interfaces and topology.
3. Layer 2 and Layer 3: VLANs, trunks, and first-hop redundancy
The user network is segmented into three VLANs, with VLAN 99 reserved as a native black-hole VLAN.
Trunks are explicitly pruned to carry only the VLANs required by each connection.
For example, SW1's uplinks carry only VLAN 10 and VLAN 99:
interface GigabitEthernet0/1
switchport mode trunk
switchport trunk native vlan 99
switchport trunk allowed vlan 10,99interface GigabitEthernet0/1
switchport mode trunk
switchport trunk native vlan 99
switchport trunk allowed vlan 10,99This limits unnecessary VLAN propagation and ensures that untagged traffic is assigned to a VLAN without user endpoints.
3.1 Inter-VLAN routing and HSRP
Inter-VLAN routing is performed by the multilayer switches through Switch Virtual Interfaces (SVIs).
HSRP provides first-hop redundancy, allowing endpoints to use a virtual gateway rather than depending on the physical address of either multilayer switch.
For example, VLAN 10 uses the virtual gateway 192.168.10.5:
interface Vlan10
ip address 192.168.10.3 255.255.255.0
standby 10 ip 192.168.10.5
standby 10 priority 150
standby 10 preemptinterface Vlan10
ip address 192.168.10.3 255.255.255.0
standby 10 ip 192.168.10.5
standby 10 priority 150
standby 10 preemptIf the active gateway becomes unavailable, the standby switch can take over without requiring any changes on the endpoints.
3.2 DHCP relay and access-port configuration
DHCP is centralized on R1.
The multilayer switches relay DHCP requests from the user VLANs using ip helper-address.
Unused access ports are administratively shut down to reduce the available attack surface.
3.3 Verifying gateway failover
I also tested whether the redundancy mechanisms behaved as intended.
By shutting down MLSW1's VLAN 10 SVI, I forced MLSW2 to assume the Active HSRP role.
The important result is that the endpoints do not need to know which physical switch currently owns the gateway role.
The virtual IP remains unchanged, demonstrating the intended first-hop redundancy behavior.
4. Routing and network services
4.1 OSPF Area 0 and equal-cost multipath routing
OSPF Area 0 handles dynamic routing between the infrastructure devices.
Loopbacks and user-facing SVIs are configured as passive interfaces. They advertise their networks without attempting to establish OSPF adjacencies.
One simulator-specific consideration is worth mentioning: Packet Tracer does not allow me to exclude a Port-channel from OSPF passivity in the same way I would prefer.
Consequently, I explicitly configured passive interfaces instead of using passive-interface default.
In a production environment, I would generally prefer the default-passive approach, explicitly enabling OSPF adjacencies only on the required transit interfaces. This makes the configuration safer when new interfaces are introduced.
4.2 Verifying ECMP toward the edge
One of the interesting aspects of this topology is the routing behavior on MLSW2.
It has two equal-cost paths toward R1:
- Through the Layer 3 EtherChannel toward MLSW1.
- Through R2 over the alternative routed connection.
As a result, OSPF installs two next-hops for edge-bound traffic.
This can be verified in the routing table, where both the default route (O*E2 0.0.0.0/0) and R1's loopback appear with two equal-cost next-hops.
This confirms that the topology provides Equal-Cost Multipath routing rather than relying exclusively on a primary path and a standby path.
It also introduces an important troubleshooting consideration: traffic may reach the same destination through different interfaces, and all relevant forwarding paths must be configured consistently.
4.3 Centralized DHCP
R1 provides DHCP services for the three user VLANs.
Infrastructure addresses are excluded from the DHCP allocation ranges, and clients receive the HSRP virtual IP as their default gateway.
For example:
ip dhcp excluded-address 192.168.20.1 192.168.20.9
ip dhcp pool VLAN20
network 192.168.20.0 255.255.255.0
default-router 192.168.20.5
dns-server 209.165.201.1ip dhcp excluded-address 192.168.20.1 192.168.20.9
ip dhcp pool VLAN20
network 192.168.20.0 255.255.255.0
default-router 192.168.20.5
dns-server 209.165.201.1
4.4 NAT/PAT and network time synchronization
R1 performs Port Address Translation (PAT), allowing multiple internal devices to share the outside interface address.
The translation rule uses an access list to identify eligible internal traffic:
ip nat inside source list NAT-ALLOWED interface GigabitEthernet0/0/1 overloadip nat inside source list NAT-ALLOWED interface GigabitEthernet0/0/1 overloadAn additional design consideration is that the NAT-ALLOWED ACL denies internal-to-internal traffic before permitting internet-bound traffic.
This documents the intended translation scope and prevents east-west traffic from becoming eligible for NAT.
It is important to distinguish this from firewall enforcement: the ACL determines which traffic is eligible for translation; it does not provide a general security boundary between internal networks.
In this topology, inter-VLAN traffic is already routed by the multilayer switches and does not traverse R1.
NTP synchronization is also configured throughout the infrastructure.
All downstream devices use R1's loopback as their internal time source, while R1 synchronizes with an external NTP server.
Consistent time synchronization is particularly important when investigating infrastructure events through centralized logging. Without synchronized clocks, correlating events across switches, routers, and security monitoring systems becomes considerably more difficult.
Although centralized log forwarding is not yet implemented in this project, the architecture provides a foundation for a future SIEM integration, including monitoring for:
- DHCP Snooping violations.
- Dynamic ARP Inspection drops.
- Port Security violations.
- BPDU Guard events.
- SSH authentication failures.
5. Edge filtering and antispoofing
The network uses an inbound ACL on R1's outside interface to filter traffic arriving from the simulated internet.
The EDGE-IN ACL drops packets arriving from outside with private source addresses belonging to 192.168.0.0/16.
Such packets should not legitimately arrive from the upstream internet and may indicate source-address spoofing or incorrectly configured upstream traffic.
For a production deployment, the filter would need to account for the full RFC 1918 private address space, alongside other appropriate ingress filtering requirements.
This is an additional security layer at the network boundary, complementing the NAT configuration.
6. The NAT incident: When every connectivity test passes, but the configuration is still wrong
This was probably the most instructive troubleshooting exercise in the entire project.
The network appeared to be working correctly.
Every VLAN could reach the internet. Pings succeeded. There were no obvious connectivity problems.
However, when I moved beyond connectivity testing and started verifying NAT translations, I noticed something unexpected.
R1 was not translating traffic from every VLAN.
6.1 Identifying the discrepancy
I started investigating the traffic paths.
As explained in Section 4, MLSW2 uses two equal-cost OSPF paths toward the edge:
- One through the EtherChannel toward MLSW1.
- Another through R2.
Traffic from VLAN 20 could therefore reach R1 through either path.
The problem was that the second inside-facing interface on R1, GigabitEthernet0/0/2, was missing the ip nat inside command.
Consequently, traffic arriving through that interface was not being processed as inside-to-outside NAT traffic.
Some flows left R1 with their original private source addresses.
6.2 Why did connectivity tests still pass?
The answer was in the simulated ISP configuration.
I had configured a static route on the ISP pointing the internal 192.168.0.0/16 address space back toward R1.
That meant packets with untranslated private source addresses could still receive replies.
In other words, my simulated environment was providing routing behavior that would not exist on a normal public internet connection.
Real internet providers do not route RFC 1918 private address space across the public internet.
The static route had effectively concealed the NAT misconfiguration.
This was a good reminder that a successful test does not necessarily prove that the underlying configuration is correct.
6.3 Verifying the correction
The resolution involved two steps.
First: Correct the NAT interface roles.
I added ip nat inside to every inside-facing interface on R1.
This ensured that traffic arriving through either of the two routing paths was correctly identified for NAT processing.
Second: Remove the artificial ISP route.
I deleted the static route pointing the ISP toward the internal private address space.
This was an important part of the validation process.
A correctly functioning NAT deployment should not depend on the ISP having a route back to private internal networks. Translated packets should arrive at the ISP using R1's outside interface address, which is already directly connected.
6.4 Testing without the artificial routing behavior
After removing the ISP's static route, I verified that outbound connectivity continued to work.
I also tested whether the ISP could initiate communication with an internal endpoint.
6.5 What this troubleshooting exercise taught me
There are three important lessons here.
1. Connectivity testing is not enough.
A successful ping proves that a packet reached its destination and a response returned. It does not necessarily prove that NAT, routing, or other intermediate mechanisms are operating as intended.
2. Redundant paths must be configured consistently.
When multiple paths lead to the same destination, every relevant interface and forwarding path must be accounted for. A configuration that works on one path may fail silently on another.
3. Test environments can conceal real-world problems.
The static route on my simulated ISP was effectively a workaround that made an incorrect configuration appear functional.
Removing that artificial behavior exposed the problem and allowed me to verify the actual NAT design.
The broader lesson is that troubleshooting should involve examining the mechanisms responsible for forwarding traffic, not just checking whether the final destination responds.
7. Defending the access layer
Coming from a blue-team background, I wanted security to be an integral part of the network design rather than something added afterward.
I implemented several access-layer controls designed to complement one another.
The objective was to protect against common Layer 2 threats, restrict unauthorized access, and generate security-relevant events that could eventually be monitored by a SOC.
7.1 DHCP Snooping
DHCP Snooping is enabled on VLANs 10, 20, and 30.
Only infrastructure-facing uplinks are trusted. Host-facing access ports remain untrusted.
This prevents unauthorized DHCP offers from being accepted through ordinary user-facing ports.
DHCP Snooping also maintains a binding table containing legitimate DHCP leases, which provides the information required by Dynamic ARP Inspection.
7.2 Dynamic ARP Inspection
Dynamic ARP Inspection (DAI) uses the DHCP Snooping binding database to validate ARP traffic arriving through untrusted interfaces.
The objective is to prevent hosts from sending unauthorized ARP messages, such as attempting to impersonate the default gateway through ARP spoofing.
The two mechanisms complement each other:
- DHCP Snooping identifies legitimate DHCP bindings.
- DAI uses those bindings to validate ARP traffic.
Without DHCP Snooping, DAI would lack its usual source of dynamic binding information.
The VLAN 20 counters show legitimate DHCP client ARP traffic being processed, with two forwarded packets and two DHCP permits.
There is, however, an important simulator limitation to document.
Packet Tracer does not fully enforce DAI binding validation.
During testing, I verified that an unbound statically addressed host could still have its ARP traffic forwarded, even with an empty DHCP Snooping binding table.
In a fully featured IOS environment, an unbound host would normally fail validation unless an appropriate static binding or ARP ACL were configured.
The controlled test in Packet Tracer therefore does not demonstrate the expected real-device drop behavior.
The simulator also reports an inconsistent VLAN 30 status in its Operation column, despite VLAN 30 not existing on this particular switch.
For this reason, I treat the configuration and available counters as evidence of the implemented inspection setup, not as proof of fully enforced DAI validation.
This distinction matters: demonstrating a security feature in a simulator is not the same as proving that every aspect of its behavior matches a production IOS implementation.
7.3 Port Security
Port Security is configured on access ports using sticky MAC address learning.
The selected violation mode is restrict.
When an unauthorized MAC address is detected, the offending frames are dropped and the violation is logged, but the interface remains operational.
This provides a balance between access-layer protection and service availability.
From a blue-team perspective, the generated event is particularly interesting.
In a production environment, this type of event could be forwarded to centralized logging and used to alert on unexpected endpoint activity.
7.4 BPDU Guard and PortFast
Host-facing ports are configured with PortFast and BPDU Guard.
PortFast allows endpoint connections to transition to forwarding without waiting through the traditional spanning-tree convergence process.
BPDU Guard protects these edge ports by placing an interface into an err-disabled state if it receives a BPDU.
This helps prevent an unauthorized switch connected to an endpoint port from influencing the spanning-tree topology.
7.5 Securing the management plane
I also implemented management-plane protections across the infrastructure devices.
The configuration includes:
- SSHv2 as the only permitted remote management protocol.
- Local user accounts with encrypted secrets.
- An access-class ACL restricting SSH access to the management subnet.
- A five-minute executive session timeout.
- Login blocking using
login block-for 120 attempts 3 within 60. - CDP disabled on host-facing ports.
Telnet is not permitted for remote management.
One simulator-specific limitation concerns password hashing.
Packet Tracer supports only legacy Type 5 hashing in this environment. Modern production platforms should use stronger supported password-hashing mechanisms.
I have documented this limitation in the project repository.
8. Where network infrastructure meets security operations
One of my main motivations for combining networking and cybersecurity was understanding how infrastructure behavior translates into security events.
This project provides a practical example of that relationship.
The controls implemented at the access layer can generate events that are relevant to security monitoring and incident investigation:
- Rogue DHCP offers: Unauthorized DHCP server activity.
- DHCP Snooping violations: Suspicious DHCP behavior.
- DAI drops: Potential ARP spoofing.
- Port Security violations: Unexpected or unauthorized MAC addresses.
- BPDU Guard err-disable events: Unauthorized switch connections.
- SSH authentication failures: Potential brute-force attempts.
These are precisely the kinds of events that could be collected and investigated through centralized logging and a SIEM such as Splunk.
My existing cybersecurity portfolio includes more than 50 practical security investigations, along with hands-on Splunk work and security monitoring exercises.
This networking project adds another dimension to that experience: understanding how the underlying infrastructure is configured, how its security mechanisms operate, and where the resulting telemetry originates.
The next step for this lab is to implement centralized syslog forwarding and build Splunk detections around these infrastructure events.
That would create a more complete connection between network engineering and security operations.
9. Validation summary
I verified the main components using the corresponding IOS show commands and controlled tests.
VLANs
- Implementation: Segmented user networks and native black-hole VLAN.
- Verification:
show vlan brief
Trunking
- Implementation: Explicitly pruned allowed VLANs.
- Verification:
show interfaces trunk
EtherChannel
- Implementation: LACP, active negotiation.
- Verification:
show etherchannel summary
Spanning Tree
- Implementation: Aligned root placement and edge protection.
- Verification:
show spanning-tree
HSRP
- Implementation: Virtual gateway redundancy and tested failover.
- Verification:
show standby brief
OSPF
- Implementation: Area 0, passive user interfaces, ECMP.
- Verification:
show ip ospf neighborandshow ip route ospf
DHCP
- Implementation: Centralized pools, relay, and address exclusions.
- Verification:
show ip dhcp binding
NAT/PAT
- Implementation: Overload, correctly designated inside interfaces.
- Verification:
show ip nat translationsandshow ip nat statistics
DHCP Snooping
- Implementation: Trusted infrastructure uplinks.
- Verification:
show ip dhcp snooping binding
Dynamic ARP Inspection
- Implementation: ARP inspection configuration and simulator validation.
- Verification:
show ip arp inspection statistics
Port Security
- Implementation: Sticky MAC learning, restrict mode.
- Verification:
show port-security interface
Management security
- Implementation: SSHv2, management ACL, login blocking.
- Verification:
show access-lists
Known limitations and future improvements
Although the lab implements the intended architecture and controls, there are several areas I would improve in a production-oriented deployment.
Dedicated management VLAN: The access switches currently use management SVIs within their respective user VLANs. A dedicated management VLAN would provide better separation.
Stronger credential protection: Packet Tracer's Type 5 hashing limitation does not reflect the stronger options available on modern platforms.
More realistic DAI validation: The simulator does not fully reproduce the binding-validation behavior expected from real IOS implementations. Testing in GNS3, EVE-NG, or physical equipment would provide stronger validation.
Centralized logging: Syslog forwarding to Splunk is a planned extension. This would allow me to correlate infrastructure events with security investigations and develop practical detections.
These limitations are documented rather than treated as completed functionality.
Conclusion: Beyond passing the CCNA
Building this project was an opportunity to move from studying individual networking technologies to understanding how they interact within a complete infrastructure.
I designed and configured a redundant collapsed-core network, implemented dynamic routing and first-hop redundancy, centralized network services, secured the access layer, and investigated a NAT issue that conventional connectivity tests failed to reveal.
More importantly, I learned to question successful test results, examine the underlying forwarding behavior, and account for the limitations of the environment in which I was working.
For me, this project represents an important step beyond earning the CCNA certification.
It demonstrates how I approach technical challenges: understand the architecture, implement a solution, verify the behavior, investigate discrepancies, and document the evidence.
The complete project is available in my [GitHub repository], including the topology, device configurations, and verification screenshots.
I welcome constructive feedback, particularly from network engineers who have experience designing and maintaining enterprise infrastructure. There is always something more to learn, and I would be interested to hear how you would approach some of the design decisions in this lab.