Tier IV audit: the 22 points your provider doesn’t want you to see

Uptime Institute Tier IV certification promises 99.995% availability, equivalent to less than 26.3 minutes of downtime per year. It is the most expensive promise in the data center market and the reason Mexican companies pay between USD $500,000 and USD $2M additional in construction versus Tier III. But between the official certificate and the site’s real operation, there are 22 points that auditors review and that providers usually overlook in the commercial pitch. If you sign a Tier IV contract without validating these 22 points with your own eyes, the risk is paying a Tier IV premium and receiving Tier III in operation.

This article delivers the 22 points grouped into 5 categories: electrical, mechanical, connectivity, operational and documentary. Each point includes what to ask, what to measure, and what provider response indicates real vs cosmetic compliance. You can do the audit yourself or with an independent consultant before signing the contract. The goal is to ensure that the money you pay corresponds to the capacity you receive.

Tier IV requires full 2N+1 redundancy and concurrent maintenance without affecting load. Tier III requires N+1 with concurrent maintenance, while Tier IV demands active fault tolerance in all systems. The difference translates into CAPEX 30 to 50% higher and OPEX 15 to 25% higher due to duplicated electrical consumption. That is why auditing before paying is critical: if your operation tolerates Tier III, there is no point in paying a Tier IV premium without real need.

Electrical infrastructure (5 points)

• Dual CFE feed physically independent: demand to see both feeds entering the site through physically separated routes. If both pass through the same CFE substation or the same trench, it is logical redundancy but not physical — it does not meet Tier IV.
• Diesel generators in 2N+1 configuration with minimum 24-hour autonomy: verify how many generators there are, their autonomy at nominal load, and whether they support peak load plus a growth margin. Ask for the date of each generator’s last maintenance.
• UPS in 2N topology with batteries sized for 15 minutes at nominal load: 2N topology means two complete and independent UPS chains, not N+1. Verify that each chain can support the full critical load without the other.
• Dual PDU with independent sources in each critical rack: each rack must have two PDUs connected to two different electrical buses, not the same one. Verify on site that the red (feed A) and blue (feed B) cables reach different buses.
• Automatic source transfer with maintenance bypass without interruption: the ATS must transfer between CFE and generator without interrupting load. Ask to see the record of the last 10 transfers and verify that the time was less than 10 seconds.

Mechanical infrastructure (5 points)

• Redundant chillers in N+1 topology with an additional chiller for maintenance: the Tier IV rule is that any chiller can fail or enter maintenance without affecting capacity. Verify there are at least N+1 chillers in continuous operation and one backup chiller.
• Dual chilled water piping with insulation and physically separated route: supply and return piping must run through different routes. If both pass through the same duct, a single hydraulic contingency takes down both routes and the system loses redundancy.
• CRAH (Computer Room Air Handler) in N+1 configuration per room, not per site: each IT room must have its own CRAH set with N+1. If the site has a single room, it is fine, but verify that the CRAHs are redundant within that room.
• FM-200 fire detection and suppression system or equivalent in each room: verify it is UL certified or equivalent, with periodic test records, sized for the actual current room volume.
• Redundant temperature and humidity sensors per room with continuous monitoring: each room must have at least 6 sensors distributed by rack density. Ask to see the dashboard and verify that data updates in real time.

Connectivity (4 points)

• Dual Meet-Me Room (MMR) physically separated: two physically independent carrier entry rooms. If carrier fibers all arrive through the same underground duct, it is not real Tier IV redundancy.
• Minimum 4 Tier-1 carriers with contracts of diverse route: two carriers is not enough for Tier IV because they share routes and POPs. Ask to see the carrier contracts and verify that entry routes are physically distinct.
• Path diversity between primary and secondary MMR verified with OTDR: ask to see the result of the OTDR (Optical Time Domain Reflectometer) that measures fibers between MMR. If fibers share ducts, there is no real path diversity.
• Redundant cross-connects between MMR with labeled cable management: verify that each cross-connect has a physical label with origin, destination, capacity and installation date. Without labeling, troubleshooting is wasted time.

Operational (4 points)

• Technical staff 24/7 on site, not just on-call: Tier IV requires staff capable of intervening at any moment. Verify there are at least 2 certified technicians on site per shift, not 1 on-call technician who arrives in 30 minutes.
• Documented concurrent maintenance procedures: ask to see the procedure manual and verify that written procedures exist for maintenance of each critical system without affecting load.
• Incident log of the last 24 months with resolution time: any serious provider has documented log. Ask to see Tier 1 incidents that affected load and verify each has root cause analysis and corrective action implemented.
• DRP plan tested at least once a year: the plan must include documented drills. A DRP on paper that has never been tested is not a real DRP — it is a document.

Contractual documentation (4 points)

• Current Tier IV certificate, not just Tier IV design: there are three Uptime certification levels (Design, Constructed, Operational). Demand the Constructed or Operational certificate, not just Design. A Design-certified DC may not yet be built or may have implementation gaps.
• Contractual SLA with compensation for non-compliance, not just reporting: if the SLA is 99.995% but the provider does not compensate monthly for non-compliance, it is cosmetic. Demand credit for real verified downtime.
• Civil liability insurance policy with verifiable coverage: ask to see the policy, not just the coverage summary. Verify coverage amount, exclusions and validity. Coverage under USD $5M is insufficient for Tier IV DC with critical load.
• Annual independent third-party audit, not just provider audit: any serious operator submits to annual third-party audit (Big Four or equivalent). If only the provider audits, you have no independent validation of Tier IV compliance.

Verdict: how to use this list

Before signing a Tier IV contract, dedicate at least 2 days to walking the site with this list in hand. The 22 points are designed so that an engineer with basic operational experience can verify them without specialized equipment beyond a flashlight, multimeter and access to the electrical room. If the provider refuses the detailed walkthrough or refuses to show documentation, assume the Tier IV certification is cosmetic and negotiate as Tier III. The CAPEX difference between Tier III and Tier IV well-justified is worth it when your load really needs 99.995%. The difference poorly justified is money thrown away on redundancy that does not operate.

Sources