Solutions for Data Center Power Path Problems

Power is as important to a data center as the heart is to a human being. Although different levels and redundancy levels can be chosen when designing a data center, no one ever wants a data center to lose power. Whether it is a small data center with only a single UPS or a large data center with full redundancy, power outages are inevitable. Here are some ways to mitigate power path problems.

Power Path: Cost vs. Acceptable Risk

Data center design is often driven by cost and acceptable risk factors. The Uptime Institute’s Tier 1-4 requirements are self-explanatory and go well beyond the topic of power, but the most basic power path is a simple N design (or Tier 1), meaning there is no redundancy and every component in the power path is a single point of failure (SPOF).

The power path begins at the generating station and its high-voltage transmission lines, then enters a substation or local transformer and is delivered to the customer. The power passes through the facility's main switchboard and subsequent sub-distribution panels, each with a circuit breaker or fuse.

Assuming there is a generator, the power must pass through the automatic transfer switch (ATS) to the main power panel that powers the UPS, which will pass through a maintenance bypass panel (MBP), and the main power panel also powers other equipment, such as the cooling system. From there, the power is sent from the UPS (or MBP) to the distribution wall panel or floor-level power distribution unit (PDU). Then it goes to the power outlet near the rack-type distribution panel, which is sometimes equipped with a circuit breaker or fuse. Finally, the power enters the power cord of the IT equipment, which is usually detachable.

It is surprising that a data center configured like this does not have power problems, but it turns out that small data centers like this are quite reliable, and many small N-type data centers can operate for many years without incident.

Increased availability: N+1 power paths

To improve the shortcomings of the N power system, a fully redundant 4-layer S+S configuration is not required. The next level of N is N+1. In the power path, it usually refers to UPS and generator sets. In N+1, there are generally three or more UPS connected in parallel. The N+1 design allows a single UPS in the UPS array to fail without affecting the power supply.

It is also possible to have a redundant AB type power distribution, where the power supply is located downstream of the UPS, and the cabinet power is supplied by a single UPS or an N+1 UPS array, thus improving the power distribution redundancy problem without the cost of a fully redundant 2N UPS system.

Finally, in a complete 2N AB system, each power path is independent to maintain the entire critical load. 2N AB is usually a 4-layer design with duplicate design for each component, two power stations, two sets of generators and related ATS equipment. Each N+1 UPS array is autonomous and provides independent AB power distribution to each rack, which is a true S+S power system.

In theory, this design allows any component to fail, with the redundant component automatically taking over for the failed component. However, in Layer 3 and Layer 4 designs, multiple intersection points between the two systems allow one to hand over power to the other in order to perform maintenance work. These connection points are often potential failure points because once crossed, the single-sided system cannot achieve full autonomy.

If a problem is caused by human error or electrical equipment failure, it could cause the entire data center to lose power. Finally, the complexity of the power path itself and the supplementary switchgear may reduce overall reliability.