Port isolation breaks the shared-screen meeting rooms the venue sells — who decides, and what do you write down?
answer
- whoever owns the revenue owns the risk
- offer graded options, not a veto
- scope it to the room, not the floor
- an exception needs an ending
- refused exceptions come back as rogue kit
basics
~20 sThe events business owns the decision, because the revenue is theirs to trade; you own stating the cost accurately. Write a per-room exception with a named owner, an expiry tied to the booking, and what one guest reaches.
solid answer
~50 sThis is not a technical call. Isolating guest ports removes guest-to-guest adjacency; the events team sells rooms where guests must reach a shared display, so the requirements genuinely conflict and the person owning the revenue is the person who can accept the risk. Your job is to make the choice a real one: offer graded options rather than a yes/no — full isolation, per-room communities scoped to a booking, or one site-wide shared-services segment — and price each in support calls, configuration effort and blast radius. Then write the exception down properly: which floor boxes, who signed, when it expires, and one plain sentence saying what a compromised guest device in that room reaches. The failure mode to design against is persistence: the event ends, the community stays, and two years later nobody can say why that room is different.
go deeper
Know that turning on port isolation has user-visible consequences and that they belong in the change plan. Expect to be asked what stops working, not just what gets safer.
Be able to state precisely which reachability an exception restores and to which ports, so the person approving it knows what changes rather than approving a feature name.
Show that you design the exception as carefully as the control: scoped to the room, owned by a named person, expiring with the booking, with the reset step assigned. Bring options and prices, not a veto.
Own the argument with the revenue owner. Be ready to hand the decision to them with the consequence in one plain sentence, to prefer a recorded exception you dislike over an unrecorded workaround, and to design so that exceptions expire by themselves.
## Why this is an ownership question, not a design question The control works. Isolation removes guest-to-guest reachability, which is exactly what you want on a floor of strangers' devices. It also removes the thing the venue is selling in its meeting rooms: a guest walks in, finds the room's display, and puts a slide deck on it. There is no configuration that gives both. When a control and a revenue line are actually incompatible, the decision belongs to whoever owns the revenue — and the security engineer's contribution is to make sure the choice is informed rather than to make the choice. The common failure is to bring it as a binary: "we need isolation on all guest ports." The events team hears "you want to break the product" and escalates, and the outcome is decided by whoever is more senior in the room rather than by the merits. Bring options with prices instead. ## Three options, honestly priced | Option | What a guest device reaches | What it costs you | |---|---|---| | Isolate every guest port | The gateway only | Shared displays and printing stop working; support calls from paid rooms | | Per-room community, expiring with the booking | Other devices in that room, for that booking | A community and a floor-box mapping per room, plus a reset discipline between events | | One shared-services segment for all displays | Every display site-wide, from any guest port | Cheapest to run; one compromised display is adjacent to a device on every floor | The middle option is usually the right recommendation for a venue, and it is worth saying why in business terms: it keeps the product intact in the rooms that are sold, confines the adjacency to people who are already in the same room together, and it expires by itself. The third option is the one that gets chosen when nobody prices it, because it is the least work. ## What actually goes in writing An exception is a security artefact, not a ticket. It needs: - **Scope in physical terms.** Which floor boxes, in which room — not "the meeting room VLAN", which will quietly grow. - **A named owner** with the authority to accept it, from the business side rather than from IT. If nobody will put their name to it, that is the answer to whether it is needed. - **An expiry** tied to something real: the booking, the contract, the next review. An exception with no end date is a permanent design decision that skipped design review. - **One sentence of consequence, in plain language.** "For the duration of this booking, any device in room 4B can reach any other device in room 4B, including a compromised one." This is what makes the acceptance meaningful; a risk nobody understood was never accepted. - **The reset step**, and who performs it. The control that matters most here is not the isolation, it is the thing that puts the room back afterwards. ## The failure you are really designing against Exceptions do not fail by being wrong when granted. They fail by outliving their reason. The event ends, the community stays, staffing changes, and the room keeps its adjacency for years — which is how an estate ends up with a rule base nobody dares touch. Every exception you grant should be cheaper to remove than it was to add, which in practice means it expires automatically or it is tied to a state the business already changes anyway, such as a booking ending. There is a related trap on the other side: refusing all exceptions produces shadow workarounds. Staff will bring a personal wireless access point into the room so the client can present, and that device is worse than anything you refused — it is unmanaged, it is on your wire, and nobody knows it exists. A recorded exception you dislike beats an unrecorded one you cannot see. ## How to close the answer Say who decides, say what you would recommend and why, say what you would write down, and say how it ends. Interviewers at this level are checking whether you can hold a security position without either surrendering it or forcing it through — and whether you understand that the durable artefact from this argument is the expiry, not the rule.
- The events director asks you to simply allow all meeting-room devices to reach each other site-wide. What is your counter?That it makes every display and every guest device on the site adjacent to each other, so one compromised device in one room is a neighbour to devices in rooms it will never be booked with. I would offer per-room scoping instead, priced in configuration effort, and let them weigh that against the operational saving they are asking for.
- How do you stop these exceptions from accumulating into a rule base nobody will touch?Tie each one to an event the business already ends — a booking, a contract term, a review date — so removal is automatic rather than a decision someone must make. Anything without such a hook gets a review date and a named owner, and an exception whose owner has left is removed by default rather than kept by default.
- What do you tell the owner about what they are accepting, in one sentence?That for the duration of the booking, any device in that room can reach any other device in that room, including a compromised one, and that the venue has no visibility of traffic between them. Acceptance without that sentence is not acceptance.
saying these in an interview costs you the question
- Treats the isolation decision as the security team's to make alone
- Grants a site-wide exception because per-room scoping is more work
- Writes the exception with no owner and no expiry
- Describes the risk in jargon the accepting owner cannot evaluate
- Refuses outright and ignores the rogue equipment that follows