Data Center Power: The Difference Between Three-Phase and Single-Phase Power

Data center architects consider current and future power requirements so they can design data centers that aren’t under- or over-powered. The use of too much power is expensive and wasteful, while the use of too little power can lead to damaging power surges when workloads spike. This is one reason why redundant power, including uninterruptible power supply (UPS) systems, cooling systems and backup generators are data center fundamentals. You know another reason: downtime is devastating.

Let’s explore data center power, including power basics and data center redundancy models and how they relate to the uptime standards associated with data center tiers.

‍Basic Data Center Power Flow

Typically, data centers are set up with several power flow elements: a power source such as a generator, UPS, power distribution unit (PDU), power panel and power whips. Depending on the size of a data center, the power flow may require more than one of the elements. 

• When a power source fails, the UPS delivers a few minutes of emergency power to provide protection against power interruptions and allow someone to properly shut down a system. 
• A PDU takes voltage from the UPS and converts it into a voltage suitable for a data center’s power panels and enables power distribution to servers, switches and other equipment.
• A power panel, which includes a breaker panel, divides the power it receives from the PDU into circuits.

Power whips connect a data center’s power panel and equipment power cords to help distribute power evenly and improve power usage effectiveness (PUE).

Single-Phase vs. Three-Phase Power

Single-phase power is called “residential voltage” because it is used in residential situations and sometimes in businesses with small workloads. Three-phase power is used in most businesses, industrial operations and data centers that have large workloads and low tolerance for downtime. Three-phase systems enable organizations to optimize PUE and uptime. A three-phase circuit provides greater power density than one-phase at the same amperage, keeping wiring size and costs lower. In addition, three-phase power makes it easier to balance loads and optimizes the utilization of electrical capacity for increase power efficiency. ²

A phase refers to the electrical current (voltage) that passes through a neutral wire. The neutral wire and a phase (live) wire deliver the power. Single-phase systems can deliver up to 230 volts of alternating current and up to 250 watts, but the amount of power varies, so delivery is inconsistent and may result in flickering lights or brief outages. Three-phase systems use three wires (or four if a neutral wire is used) and alternating currents to generate 415 volts. Unlike the single-phase systems, the power output in three-phase systems is continuous and consistent. 

As an aside, keep in mind that direct-current (or DC power) is preferred to alternating-current (or AC power) in data centers to eliminate power conversion steps and losses, reduce cooling requirements and support equipment density. 

Data Center Redundant Power Systems

Redundant power systems minimize the risk of unplanned downtime and, in turn, the negative consequences. In data centers, key redundancy components are: 

• UPS systems, in which lithium-ion batteries are increasingly used to store power.
• Cooling systems such as chillers with magnetically levitated centrifugal compressors (also known as maglev systems) and free outside air that may pass through an evaporative cooler before it enters a data center.
• Backup generators, which switch into action when the power source is interrupted for any reason. 

The industry has standardized on certain redundancy levels that are a useful guide to understanding how an on-premises or colocation data center is designed. 

• N equals the amount of capacity required to power, back up or cool a facility at full IT load.
• N+1 or N+X redundancy adds a single component (or X components) – a UPS, HVAC system or generator – to each element of the N architecture to support a failure and maintain a full workload.
• 2N redundancy creates a mirror image of the original UPS, cooling system or generators to provide full fault tolerance.
• 2N+1 delivers the fully fault-tolerant 2N model plus an extra component to each element of the N architecture for extra protection to withstand multiple component failures. Even when the primary system is offline, N+1 redundancy is sustained.
• 3N/2 adds additional capacity based on the load of the system. For example, three power delivery systems provide power to two servers.

Redundancy and Data Center Tiers

Data center classification tiers describe the infrastructure and redundant power components. Let’s say you’re evaluating colocation providers. Your business requirements related to risk tolerance and cost are top factors in deciding which tier is right for your organization. The tiers, according to Hewlett-Packard Enterprise, are:

• A Tier 1 data center has a single path for power and cooling and few, if any, redundant and backup components. It has an expected uptime of 99.671% (28.8 hours of downtime annually).
• A Tier 2 data center has a single path for power and cooling and some redundant and backup components. It has an expected uptime of 99.741% (22 hours of downtime annually).
• A Tier 3 data center has multiple paths for power and cooling and systems in place to update and maintain it without taking it offline. It has an expected uptime of 99.982% (1.6 hours of downtime annually).
• A Tier 4 data center is built to be completely fault tolerant and has redundancy for every component. It has an expected uptime of 99.995% (26.3 minutes of downtime annually).

Power and Redundancy Support Uptime

On-premises and colocation data centers require reliable uninterrupted power. Surprise outages can be disruptive and expensive. According to the Uptime Institute ³:

• Over 60% of failures result in at least $100,000 in total losses, up substantially from 39% in 2019.
• The share of outages that cost upwards of $1 million increased from 11% to 15% over that same period.
• Nearly 30% of these [major public] outages in 2021 lasted more than 24 hours, a disturbing increase from just 8% in 2017.

Want to explore a data center and its infrastructure? Book a tour and bring your questions about power requirements. And learn more about uptime and things to think about related to data center tiers in this CoreSite blog.