Exploring the Benefits of Redundant UPS Systems in Data Centers

There are a variety of UPS systems and redundancy configurations to choose from. Selecting the right one for a given environment depends on the reliability and fault tolerance required.

A 2N redundant architecture delivers the highest levels of reliability. However, this can increase upfront and operational costs and may require a larger installation floor space.

Reliability

An essential element of any data center UPS system is its ability to provide backup power for a short period when mains power fails. This provides time to shut down equipment safely or activate the backup generators.

Using multiple UPS units in parallel increases redundancy and enables data centers to achieve higher availability and MTBF (mean time between failure) ratings. It also allows maintenance to be conducted without interrupting the critical load.

A common form of redundant UPS is the hot standby configuration. With this design, the normal system power flow to the critical load passes through UPS #1, with UPS #2 acting as a standby. If either UPS experiences a fault, the static bypass will transfer the load to the secondary unit within 4-5 milliseconds.

Another option is the distributed redundant UPS system. This configuration utilizes two independent UPS systems to support the critical load, and a third system is installed to serve as a distributed bypass. Combined with a static transfer switch, this design can provide an N+2 redundant power solution for the critical load without using a single point of failure.

Another popular redundant UPS solution is the series redundant configuration. This configuration reduces cost by using two or more UPS units with a common battery bank. Each UPS has a run time of half the required capacity, and, in the event of a failure, the first unit will transfer to bypass, and the second will take over the full load.

Availability

Data center equipment relies on UPS systems to keep it running during power outages. When a power disruption occurs, the UPS system provides temporary battery backup that keeps critical equipment running until normal power can be restored. A reliable UPS system allows data centers to retain valuable information and maintain business operations.

Five primary UPS systems design configurations distribute power from the utility source to the data center’s critical loads. Each has its impact on availability, and the configuration chosen should be based on the data center’s availability needs, risk tolerance, type of load, budgets, and existing infrastructure.

A standard N+1 redundancy configuration provides the lowest cost and simplest approach, requiring one additional component for every four required to support full capacity. In contrast, an N+2 redundancy design adds two more components for an added level of protection but may require complex and precise load planning.

A distributed redundant configuration (DRU) is a popular option for larger multi-megawatt installations where concurrent maintenance and limited space are needed. A DRU features multiple UPS modules that are paralleled and sized to match the critical load projection. The paralleled UPS modules share the critical load equally during normal operation, and if a module fails, the remaining modules transfer to straight utility power. In this way, the DRU delivers 2N reliability with N+1 capital and operating costs.

Efficiency

When a UPS system fails in data centers, it can cause hardware or file system damage. Even a short power outage can result in lost data and downtime for thousands of users logged into a system. Fortunately, redundant UPS systems allow for more uptime during a power disruption.

A redundant UPS system operates with a modular design and is easily scalable, with the capacity to meet growing IT needs. These systems also offer redundancy configurations and dual bus capabilities to ensure your critical equipment will remain online during blackouts, brownouts, sags, or surges.

Many facilities use N+1 plans to power, backup, and cool their IT equipment. N represents the number of components needed to run a facility at full IT load, and the plus 1 adds additional capacity to handle an individual component failure or required maintenance.

Distributed redundant configurations provide N reliability and can increase the power output of a system while reducing capital and operating costs. This type of architecture flows power from the utility into two separate UPS modules/PDUs connected to the server environment. The modules work together and share 80% of their capability. If one UPS module fails, the other takes over the load. This differs from the parallel redundant design, often called N+1.

Cost

As data centers grow, they require more power and higher backup capabilities. To avoid costly disruptions to production operations during an extended outage, some facilities opt for redundant UPS systems that offer multiple redundancy configurations, dual bus capabilities, and more.

The cost of redundant UPS systems depends on the required redundancy level and the configuration type used. Distributed redundant configurations, more common in large data centers today, use a series of UPS modules that are grouped together with independent input and output buses connected to the critical load. This allows a single module to fail without disrupting the critical load or forcing it back to the utility source.

Another option is a parallel redundant design, which uses two sets of UPS modules synchronized by an internal or external paralleling function. The modules do not need to be of the same capacity or even from the same manufacturer, but they must be able to accept and share the critical load in normal operation.

Choosing the right redundant UPS system requires careful consideration of how many servers or other equipment the facility supports and the capacity requirements for voltage, current, and power inrush. Many older UPS systems need to be bigger, which drives up electricity costs and leads to wasted capacity. Using newer UPS technology with high efficiency and redundant configurations can save thousands of dollars in running costs while increasing availability.