Correct Answer:

See The answer in Explanation part below.
Explanation:
Step 1 - Confirm the immediate tester-level response was correct. Stopping all interactive activity with the application the moment anomalous behaviour was observed (19:54) was the right first action, consistent with the RoE's implicit expectation that testers exercise caution around any sign of potential service impact, even absent an explicit instruction to halt at that exact moment. This should be affirmed, not criticised, in any post- incident review - the tester exercised appropriate professional judgement.
Step 2 - Recognise the escalation channel has failed, and escalate further immediately. The named stop- testing contact being unreachable by both listed channels is a serious, live risk-management gap: the RoE's single point of contact and single backup channel have both failed simultaneously. The tester (and you, once informed) must not simply wait passively. The correct immediate action is to escalate through any other reasonable, available means: contacting the Control Group chair or other known senior client stakeholders directly (even if not the named RoE contact), using any other documented emergency contact details held by your firm (e.g., from the kickoff meeting contact list, main switchboard, or account management relationship), and internally escalating to your own firm's senior management/Test Director so the incident is being actively managed rather than left with a single tester.
Step 3 - Preserve evidence and document a precise timeline. You and the tester should immediately and precisely document the timeline: exact timestamps of the observed anomaly, the decision to stop, and every attempted escalation contact (including the voicemail and bounce-back), together with exactly what technical activity was being performed in the minutes before the anomaly appeared. This record is essential both for genuinely understanding whether the Red Team's activity contributed to the issue, and as a contemporaneous account protecting the firm and the individual tester if the legality or conduct of the engagement is later questioned.
Step 4 - Do not resume testing on the affected system until contact and clarity are achieved. Testing on the affected application (and arguably more broadly, pending clarification) should remain paused until the Red Team Manager has made actual contact with an appropriate, accountable client stakeholder, confirmed the client's current understanding of the system's status, and received explicit direction on whether and how testing should continue. Resuming activity on the affected system without this confirmation, simply because the scheduled window reopens the next morning, would be an unacceptable risk given the unresolved uncertainty about what caused the slowdown.
Step 5 - Once contact is made, support the client's own investigation. When a client contact is finally reached (whether that evening or the next morning), the Red Team Manager should proactively share the precise timeline and technical detail from Step 3, to help the client's own team determine quickly whether the Red Team's activity was a contributing factor, and offer to pause the wider engagement if needed while this is established, rather than downplaying the incident to keep the schedule on track.
Step 6 - Identify and remediate the governance/RoE weaknesses exposed. Before testing resumes, several weaknesses must be addressed and, where appropriate, formally reflected in an updated RoE through change control: (i) reliance on a single named individual with no genuinely independent backup contact is a single point of failure and should be replaced with at least one alternate/deputy contact with equivalent authority, consistent with the continuity planning principles covered elsewhere in the syllabus; (ii) the backup email channel being allowed to reach a full, non-monitored mailbox indicates the channel was not actually being maintained as a reliable emergency channel - this should be tested/verified periodically, not merely documented on paper; (iii) the incident should prompt a rehearsal or "dry run" check of the stop-procedure contacts going forward, consistent with the syllabus principle that escalation procedures benefit from practical verification, not just written definition; and (iv) the Control Group should be briefed on the incident and the contact/process gaps, so it can decide on any wider corrective action.
Conclusion: The tester's decision to halt activity was correct and should be reinforced; the priority afterward is aggressive, multi-channel escalation and evidence preservation rather than passive waiting or unilateral resumption; and the incident should trigger a formal review and strengthening of the RoE's single-point-of- failure escalation contact structure before testing continues.
---

Correct Answer:

See The answer in Explanation part below.
Explanation:
Step 1 - Recognise the shared-facility authorisation problem. The data centre facility itself, and the general access points, common areas, and physical security controls governing entry to the building, are owned and operated by the data centre operator - a separate legal entity - not by Silverpeak. Silverpeak's authorisation can validly cover its own locked cage and the equipment within it, but it cannot validly authorise a physical intrusion attempt against the building's general access controls, which are the data centre operator's own infrastructure and responsibility, exactly analogous to the cloud/SaaS/telecommunications-provider authorisation-boundary issue addressed elsewhere in this syllabus, now applied to a physical rather than purely technical context.
Step 2 - Recognise the additional multi-tenant risk dimension. Beyond the pure authorisation question, a physical intrusion attempt against the shared facility risks affecting or alarming the other two unrelated Authorized Institutions whose cages are in immediate physical proximity - for example, if the attempt triggers a wider facility security response, lockdown, or law enforcement involvement affecting the whole building, not just Silverpeak's area. This mirrors the "shared multi-tenant environment" risk principle covered elsewhere in this syllabus regarding cloud infrastructure, now applied physically, and materially raises the stakes of proceeding without the operator's explicit involvement.
Step 3 - Do not proceed with the physical facility-entry component as currently framed. Given Steps 1 and
2, this specific element should not proceed on the basis of Silverpeak's authorisation alone. The professionally correct response to the Control Group is to explain clearly why their own authorisation cannot legally or safely extend to testing the shared building's general access controls, however enthusiastic they are about a comprehensive test.
Step 4 - Identify legitimate alternative approaches. Rather than simply declining outright, you should discuss constructive alternatives with the Control Group: (i) engaging the data centre operator directly to seek their explicit, separate consent for a properly scoped and coordinated physical test of the building's general access controls (which, if obtained, would need to be documented and would still require care given the other tenants' interests, potentially requiring their awareness or at least the operator's confirmation that testing is compatible with its own obligations to other tenants); (ii) narrowing the physical testing component to elements genuinely within Silverpeak's own control, such as testing access controls on Silverpeak's own locked cage itself (e.g., attempting to gain entry to the cage assuming a tester has already reached the general shared area through legitimate means, or testing whether Silverpeak's own escort/visitor procedures are followed by data centre staff who do have authorised access) - carefully scoped to avoid implicating the operator's own general building security; or (iii) excluding physical facility testing from this engagement and instead documenting physical access risk at the shared facility as a topic for Silverpeak's own vendor/facilities risk management and direct conversation with the data centre operator outside the iCAST engagement itself.
Step 5 - Address the legal-position concern rigorously, not informally. The team member's instinct that Hong Kong's legal position may differ from a "UK-only" assumption is exactly correct as a concern, and it should not be left informally unresolved. Consistent with the syllabus principle on jurisdiction-specific legal risk, your firm should not proceed with any physical social engineering element in Hong Kong based on assumptions carried over from UK engagements. This requires confirming (through your firm's own established Hong Kong legal understanding, given this is an iCAST-accredited engagement where such understanding should already exist, or through specific local legal advice if any doubt remains) the local legal position on trespass and physical intrusion testing, and ensuring the authorisation and RoE documentation for this specific engagement explicitly and correctly reflect that position, rather than being inherited unreviewed from unrelated prior UK engagements.
Step 6 - Document the resolution and rationale. Whatever combination of Steps 4's alternatives is ultimately agreed with the Control Group, the rationale, the authorisation boundary reasoning, and the confirmed legal position should be clearly documented in the engagement's scope and RoE documentation, both for internal audit trail purposes and to support any eventual C-RAF/HKMA-related review of the engagement's conduct.
Conclusion: The shared data centre's general building access controls cannot be validly authorised for testing by Silverpeak alone and should not be included without the data centre operator's own explicit, separately obtained consent, given both the authorisation-boundary principle and the added risk to unrelated co-tenants; and the team's informal, unverified assumption about Hong Kong's legal position must be properly and specifically confirmed (not carried over from UK experience) before any physical social engineering proceeds.
---