Rated-3 claim with a single PDU and a common cooling loop
A financial services firm claims Rated-3 for its 1.2 MW data centre. The electrical design is 2N UPS, but each rack has a single row PDU and the dual-corded servers have both cords landed on that same PDU; several SAN switches are single-corded. The UPS maintenance procedure requires transferring the load to the bypass, which causes a planned 10-minute IT interruption. Cooling is N+1 CRAH units on a common chilled-water loop with no isolation valves; maintaining a pump requires shutting down the loop. The site has a maintenance window every quarter, and operations says '2N UPS means we are Rated-3'. The audit criteria are the applicable edition of TIA-942 and the site's own Rated-3 claim. You are reviewing the claim for the annual internal audit.
1. Assess whether the Rated-3 claim is supportable. Identify the specific TIA-942 concepts that are not met.
Read the worked answer
Rated-3 means concurrently maintainable: planned maintenance on any capacity component or distribution path should be possible without affecting the IT load. A 2N UPS configuration is relevant but not sufficient. The rating depends on the entire power and cooling path, including distribution to the rack and the procedures used.
The rack topology fails the concept. If both cords of a dual-corded server land on the same row PDU, that PDU is a single point of failure for maintenance and for faults. Maintaining or replacing the PDU, or the upstream UPS output circuit, requires the load to be transferred or shut down. Single-corded SAN switches are worse: they cannot be maintained without interrupting the SAN path unless an alternate path and rack-level transfer arrangement exist. A 10-minute planned interruption in the UPS maintenance procedure directly contradicts concurrent maintainability; the procedure is evidence that the design cannot meet the claim as operated.
The cooling system also fails. N+1 CRAH units provide redundant capacity components, but a common chilled-water loop with no isolation valves means pump or loop maintenance requires shutting down the loop. That is a planned outage of cooling to the IT load, not concurrent maintenance. The site may have redundant components, which is closer to Rated-2, but it has not demonstrated Rated-3. The correct conclusion is that the Rated-3 claim is not supportable for the whole site as configured and operated.
2. What evidence and audit techniques would you use to confirm the gap and avoid over-relying on interviews?
Read the worked answer
Start with documents that define the as-built and as-operated state: single-line diagrams, rack power topology drawings, PDU and ATS/STS ratings, cooling schematics, valve schedules, and the maintenance procedures (MOPs) for UPS, PDU and chilled-water pump work. The MOP that states a 10-minute interruption is strong documentary evidence because it is the organisation's own instruction for planned work.
Use observation to verify that drawings match reality. Walk sample racks across each hall and record whether dual-corded equipment has cords on separate PDUs or the same PDU, and whether single-corded equipment has a rack ATS. Inspect the chilled-water loop for isolation valves at pumps and CRAHs. Ask for the last actual maintenance record and compare what was done with the MOP. If a UPS maintenance event occurred, examine the switching sequence, the load readings and any IT incident ticket raised during the window.
Interviews should corroborate, not lead. Ask the facility engineer to walk through the UPS bypass sequence and the pump maintenance sequence, and ask IT operations whether any planned work required an application shutdown. Sample CMMS work orders for the last 12 months and look for outage approvals, bypass permits and change records. Where possible, witness or review a simulated transfer test. Triangulate documents, observation and records; a single interview statement that 'it is all redundant' is not sufficient.
3. What rating would you defend in the report, and what remediation priorities would you set?
Read the worked answer
I would not defend a whole-site Rated-3. I would report the rating by system and state the basis. Electrical distribution to the rack is not concurrently maintainable because of the single PDU and single-corded SAN switches; cooling is not concurrently maintainable because the common loop cannot be isolated. A defensible interim statement is that the site demonstrates some redundant capacity components but does not demonstrate Rated-3, with the overall operating rating limited to Rated-2 until the gaps are corrected and verified. The report should make clear that this is a conformance conclusion against the stated claim, not a redesign opinion.
Remediation priorities should follow risk and dependency. First, correct the rack power topology: dual PDUs on separate UPS paths, or listed rack ATS for single-corded equipment, with a documented verification that either path can be maintained without load loss. Second, add isolation valves and a pump/CRAH maintenance procedure that allows concurrent work, or provide an independent cooling path. Third, rewrite the UPS and PDU MOPs so planned maintenance does not require an IT interruption, and test the sequence. Fourth, update the rating claim and customer documentation immediately, because the current claim is public and contractual.
Re-audit after remediation with the same sampling method and require evidence of a successful maintenance event performed without IT impact. Until then, record the claim as not demonstrated.