Data centers are critical infrastructures that support business, government, and defense systems and deliver smooth online services to users. However, data centers are also extremely power-hungry and create intense microclimatic conditions through the tremendous heat generated from their server racks, which must be constantly cooled. Additionally, a very specific temperature range (64°F - 80°F) needs to be maintained within the premises.
Data centers are designed to run 24/7/365, and even a slight drop in power can have significant consequences, such as:
Servers overheating
Loss of service provision
Data corruption and loss
Destruction of equipment
Fire danger in the facility
Any server downtime will hurt data center customers. Even if an individual customer isn’t too bothered, you can guarantee an ISP or streaming service will not be happy if they're offline for even a few minutes. Maintaining uninterrupted operations at data centers is a responsibility to customers and investors and, ultimately, an issue of revenue security, protecting the reputation and equipment of the data center operator.
The Role of Critical Backup Power in Data Centers
In this blog, we’ll look at the role of critical backup power in data centers and how it ensures continuous service and revenue security.
Backup Generators
Emergency diesel generators have been providing backup power for over a century. The modern emergency genset can be manually, remotely, or automatically operated, has programmable functionality, and can cover the data center’s entire needs or just predetermined critical loads. Diesel and gas-fueled generators can provide a comparable energy output to mains supply, provided they are installed to the necessary capacity for the data center’s requirements.
Depending on the size of the genset, it is possible to run continuously for days or weeks on the assumption that fuel can be provided, but even standard storage will entail a diesel generator having a 24, 48, or 72-hour fuel tank. For propane or other natural gas generators, fuel storage can be higher as the gas can be compressed into a smaller space, which is important since most data centers try to maximize their servers per square foot.
However, emergency gensets need time to start and pick up the required load. Additionally, maintenance and testing are required regularly to ensure they are working as intended and don’t fail when needed.
Uninterruptible Power Supply
Uninterruptible power supply (UPS) systems are designed to instantly pick up critical loads in a data center without uptime. These systems also ensure that even minor outages or fluctuations in power are smoothed over, minimizing the risk of equipment damage and reducing the reliance on quick ignition for backup generators. Some common UPS units used in data centers include large batteries, supercapacitors, and flywheels (as part of a dynamic, uninterruptible power supply).
UPS systems are used in many tech environments where even a moment of lost mains power could result in corrupted or lost data. However, for data centers, the units must be much larger to carry the high loads involved, even if only for a few seconds or minutes. A data center’s UPS system is usually also intrinsically linked with its power distribution and HVAC system, allowing greater optimization of power usage, such as breaking the facility into discrete sections so temperature levels can be maintained on a section-by-section basis.
An uninterruptible power supply will usually be measured in minutes as battery or other energy storage technology doesn’t yet exist which can match the energy levels of diesel or gas in similar volumes. Loads will be switched to the emergency genset when it comes online.
Built-in Redundancy
Permanent uptime data center reliability and the resultant revenue security require planned emergency generation redundancy. These are implemented in arrangements such as N, N+1, N+2, 2N, and 2N+1, where N is the minimum capacity required to power a data center at full load, and N+1 means there is one extra component that can provide power in the case of multiple component failures.
For example, N is the most basic redundancy architecture, meaning if the data center requires six UPS battery units in the case of mains failure then that’s all you have installed, N+1 would mean having seven units in case one fails, while 2N would mean having twelve units. With costs of data center downtime running up to $5 million hourly, having adequate redundancy in place to cover an eventuality that sees multiple simultaneous component failures secures your business’ reputation and helps prevent or minimize losses.
Conclusion
Data center downtime can have major consequences, including data loss or corruption, equipment damage, and reputation loss, even resulting in litigation or triggering contractual compensation. There is an expectation that data centers will run 24/7/365 with next to zero downtime. To ensure this a data center must have sufficient backup power, formed of emergency gensets and intermediary UPS systems, which cover the transition period.
A data center must also be designed to include redundancy for that backup power in the case of component failure or destruction. These elements are necessary to minimize potential downtime and secure your organization’s reputation and revenue.
At Enercon, we have helped design, build, and install critical backup power systems in a wide range of civil, defense, and industrial fields, including data centers. To learn more about how we can help you minimize downtime and optimize performance at your facility, contact us today.