VLANs, Multihomed Hosts, and Segmentation Evidence
An assessment diagram shows a user network, a management network, and a server network. They’re drawn as separate zones, so it’s tempting to conclude that traffic can’t move directly between them. Then you discover a management server with one network interface in the user zone and another in the management zone.
Have you proved a segmentation bypass? Not yet. You’ve found a connection that deserves investigation, but the diagram and the interface list don’t tell you whether the host forwards traffic, whether its firewall permits that traffic, or whether the connection is an intended and authorized part of the design.
That difference between observing connectivity and proving a security weakness is central to this lesson. CompTIA PenTest+ objective 4.2 includes VLAN hopping, multihomed hosts, relay attacks, share enumeration, and packet crafting among its network attack topics. Those labels identify relevant technique families. They don’t prove that any particular network is vulnerable, and they don’t replace the need to understand the switching, routing, authentication, and trust configuration in front of you.
A virtual local area network, usually called a VLAN, lets a switching design separate traffic into logical network segments. VLAN hopping is an attempt to send or receive traffic across those boundaries in a way the design didn’t intend. The important words are didn’t intend.
If a router or other authorized network device forwards permitted traffic between two VLANs, that’s ordinary network connectivity governed by the design. If an administrator’s workstation is deliberately connected to two management environments for an approved purpose, two interfaces alone don’t make it an attack path. A finding requires evidence that an intended boundary can be crossed contrary to the relevant policy or configuration.
The same discipline applies to a multihomed host, which is a host connected to more than one network. The second interface may create potential reachability, but potential isn’t the same as forwarding capability. The operating system’s routing behavior matters.
Host firewall rules matter. The services listening on each interface matter. Authentication and trust relationships may matter.
A host might communicate separately with both networks while refusing to carry traffic from one to the other. It might also provide an approved application that accepts a request on one side and communicates with a service on the other. That’s multi-network connectivity, but it isn’t automatically an unauthorized pivot.
Relay attacks involve forwarding or reusing an authentication exchange or another protocol message toward a different endpoint. Here again, the technique depends on prerequisites. The relevant endpoints must be reachable, the protocol exchange must be suitable for relaying, and the receiving service must accept the resulting exchange under its authentication and trust configuration.
Merely seeing authentication traffic doesn’t establish that it can be relayed successfully. Nor does an attempted relay prove access if the destination rejects it. You need endpoint evidence showing what received the traffic and what consequence followed.
Shares require similar care. Share enumeration means identifying shared resources and the information exposed about them. A share’s existence isn’t the same as unauthorized access to its contents. Connectivity, authentication, share permissions, and the surrounding trust configuration affect what a particular identity can discover, read, or change.
An assessment should distinguish a publicly visible share name, authorized read access, excessive read access, and unauthorized write access. Those are materially different observations with different consequences. The safest validation is the one that establishes the relevant permission without unnecessarily opening, copying, changing, or deleting sensitive material.
Packet crafting is the deliberate construction or modification of network packets for a testing purpose. It can help examine how a device or service handles traffic with selected protocol characteristics. But a crafted packet is an input, not a finding.
Sending it doesn’t tell you by itself whether the intended endpoint received it, whether an intermediate device changed or blocked it, or whether the destination accepted it. That’s why protocol-level packet evidence matters. A capture analyzed with Wireshark can help validate whether crafted or relayed traffic reached the intended endpoint and whether the observed exchange matches the prediction.
Before testing any of these paths, state the expected result. Which device should receive the traffic? Through which interfaces and network boundaries should it travel? What response would support the hypothesis?
What result would weaken it? This prediction prevents a vague success message or an isolated packet from carrying more meaning than it deserves. It also helps you choose a bounded action rather than reaching immediately for the most intrusive technique available.
Return to the dual-homed management server. One interface is connected to the user subnet, and the other is connected to the management subnet. The approved engagement permits validation of segmentation between those two networks during a defined maintenance window.
The first observation is that the server itself can communicate with systems on both subnets. That proves the server has connectivity to each side. It doesn’t prove that a device on the user subnet can use the server to reach the management subnet.
Your initial hypothesis might be that forwarding through the server is blocked. If that hypothesis is correct, traffic presented from the user side for a management destination shouldn’t emerge from the management-facing interface. A bounded validation can compare captures or other approved evidence at the relevant points. If traffic appears on the user side but never appears on the management side, the result weakens the claim that the host is forwarding it.
The reason could involve routing, the host firewall, another configuration control, or a flaw in the test setup. The evidence tells you where the observed path ended. It doesn’t justify guessing which control caused that result.
Suppose instead that matching traffic appears at the management-side observation point and reaches the intended management endpoint. That’s stronger evidence that the path crossed the expected boundary. Even then, you must compare the path with the authorized design.
The management server may be an approved gateway or application intermediary. If its documented purpose includes that communication, the observation may confirm intended routing rather than reveal a bypass. To support a segmentation finding, you need the technical path and the policy context showing that the source, destination, protocol, or direction wasn’t supposed to be allowed.
This is why topology evidence has to describe actual connections rather than merely repeat network names. Record the source device, the interfaces involved, the expected boundary, the destination, and any authorized intermediary. Then align packet evidence with that path. Addresses and protocol exchanges can help distinguish traffic sent directly to a destination from traffic sent to an intermediary.
Capture locations matter too. Seeing a packet near the source proves it existed there. It doesn’t prove that the packet crossed every later boundary or reached the final endpoint.
Timing matters as well. Evidence collected before a configuration change may no longer describe the tested state. A capture from the wrong interface may omit the relevant traffic. A destination selected from an outdated diagram may not represent the boundary you intended to examine.
When observation and prediction disagree, first revisit target selection, capture placement, configuration, privileges, prerequisites, and data freshness. A failed observation can mean that a control worked, but it can also mean the assessment looked in the wrong place. That uncertainty should remain visible until corroborating evidence resolves it.
Consider a relay test in an authorized lab. You observe an authentication exchange and predict that a receiving service might accept a relayed exchange. A packet capture near the first endpoint can show the original protocol traffic. Another observation near the intended receiving service can show whether related traffic arrived there.
The receiving service’s response can then indicate whether it rejected or accepted the exchange. These pieces answer different questions. Original traffic proves that an exchange occurred.
Arrival evidence supports that it was forwarded. An acceptance response supports a consequence. Treating the first piece as proof of all three would overstate the result.
A similar distinction protects you when evaluating shares. Enumeration might reveal a share name and some descriptive information. That’s evidence of discoverability, not necessarily evidence that files are readable. If the engagement authorizes permission validation, use the least-impactful method that can establish the relevant access level.
Don’t turn a question about listing or reading into an unnecessary write test. If stronger validation would touch sensitive data or alter a resource, authorization, target scope, timing, and operational constraints should decide whether that action is permitted. Technical curiosity doesn’t silently expand the engagement.
Operational impact is especially important around management systems and operational technology. A network technique that seems routine in an office environment may carry different reliability or safety consequences in another environment. You still need evidence, but the method used to obtain it must fit the operational constraints. If packet captures, existing logs, configuration review, or another lower-impact observation can establish the path, use that evidence before escalating. If the approved test can’t establish the condition safely, document the limitation instead of forcing a more disruptive action.
Now consider conflicting evidence from the dual-homed server. A network scan from the user subnet reports a management-side address as responsive, but the capture on the server’s management interface shows none of the expected test traffic. That scan result shouldn’t immediately become a pivot finding.
Confirm that the responding address belongs to the intended destination. Check whether another authorized router provides a path. Verify that the capture covered the right interface and time window.
Determine whether cached, stale, or misidentified data affected the observation. A tool result is an observation. The finding comes from explaining the path and validating that explanation.
The high-risk misconception is simple: two network interfaces equal a usable pivot. They don’t. To act as a pivot or segmentation bypass, the host must enable some usable movement between the networks, and the surrounding controls must permit the relevant traffic or protocol behavior. A routing table or interface listing can support your understanding of the host, but neither alone proves end-to-end movement. The verification step is to look for authorized, correlated evidence that traffic from the source side crossed through the relevant path and reached the intended destination side.
You should apply the same standard to VLAN hopping. A packet that contains a chosen characteristic isn’t proof that switching behavior carried it into another VLAN. Capture evidence near the sender shows what was transmitted. Evidence associated with the intended destination side helps show whether it arrived.
A response may provide additional corroboration. Then the switch design and approved topology tell you whether the observed path violated segmentation or followed an allowed connection. The technique name gives you a hypothesis. The network’s behavior supplies the evidence.
Packet crafting supports this process when it gives the test a precise, recognizable input. You can predict what an accepted, blocked, or altered exchange should look like, then compare that prediction with captures from approved observation points. The interpretation still has to remain modest.
No response doesn’t automatically mean a packet was blocked at the boundary. It might have reached a silent service, followed an unexpected route, or been missed by the observation setup. Conversely, a response proves some form of communication, but the addresses, protocol details, capture points, and topology must establish which boundary was crossed.
Imagine you must choose between two next actions. One choice is to launch a broader relay or packet-crafting attempt across every visible segment. The other is to verify the approved source and destination, examine the topology, predict the expected exchange, and collect bounded packet evidence during the authorized window. The second choice is defensible because it tests a specific mechanism while controlling scope and impact. The broader attempt may produce more output, but more output isn’t better evidence when the targets, trust assumptions, and operational effects aren’t understood.
The answer could change if the engagement explicitly authorizes wider testing and the lower-impact evidence can’t resolve an important question. Escalation may then be justified, but it should be deliberate. State what remains unknown, why the next action can answer it, which systems are included, and what operational limits apply. A technically valid technique used against an excluded target or outside the approved time is still the wrong decision. Authorization is part of correctness, not an administrative detail added after the technical work.
Documentation should preserve the reasoning chain. Record the expected boundary, the relevant source and destination, the actual interfaces or network paths, the trust or authentication assumption being tested, and the evidence obtained. Separate observed facts from interpretation.
For example, you might report that matching protocol traffic was observed entering and leaving a dual-homed host during an authorized test, and that the intended destination responded. Then explain whether the documented design allowed that path. If the design information is incomplete, say that the security significance remains unresolved rather than presenting uncertainty as a confirmed bypass.
A useful report also distinguishes the affected control from the technique used to test it. VLAN hopping, relaying, share enumeration, and packet crafting describe activities or technique families. Remediation has to address the actual weakness supported by the evidence, such as an unintended network path, an inappropriate trust relationship, or access that exceeds the intended design. If the test only established discoverability and not access, the report shouldn’t claim access. If it proved reachability but not unauthorized forwarding, it shouldn’t call the host a successful pivot.
This approach leaves the next person with something reproducible and reviewable. They can identify what was authorized, what state was predicted, what bounded action occurred, where the evidence was collected, and which conclusion that evidence supports. They can also see what wasn’t established. That matters because network boundaries aren’t proven by lines on a diagram, and they aren’t disproven by a technique name. They’re understood by connecting configuration, topology, protocol behavior, and policy to the traffic that actually crossed the network.
VLAN hopping, multihomed hosts, relays, shares, and crafted packets all raise the same practical question: what did the network actually permit, and was that behavior intended? An extra interface, a visible share, or a transmitted packet is only an observation. A defensible finding connects that observation to the real switching, routing, authentication, and trust configuration, then corroborates the path with appropriate endpoint, topology, and packet evidence. Keep the validation bounded by the authorized targets, time, and operational limits, and use a lower-impact method when it can answer the question. When you can separate intended connectivity from an unauthorized boundary crossing and document the remaining uncertainty, you’re making a technically sound assessment rather than merely collecting attack output.
Pause and decide: Have you proved a segmentation bypass? Consult the companion episode reference for exact terminology, named tools or standards, source-derived evidence checkpoints, and review material for Segmentation Bypass, Relays, and Packet Crafting. This lesson was produced by Bare Metal Cyber dot com.