"A constant power supply is the basic requirement of the data center. Without sufficient, uninterruptible energy, the complex framework that stores information and provides network support is rendered moot.
As enterprises in many industries across the world enhance the scope of their data center outsourcing practices, power is increasingly pulled into the limelight. Cloud computing, colocation and disaster recovery planning are just a few of the major phenomena driving businesses to link their fortunes to data centers. With more at stake, power availability is critical.
Since energy availability and power consumption have consequences for nearly every aspect of data center functionality, heightened focus on them is largely a positive thing. Across the globe, some of the top technological and business innovators are investing in ways to improve energy efficiency, an initiative that may have positive implications for the environment as well as for enterprise wallets.
Like any rapidly transforming field, data center power demands can be difficult to keep up with. Developments in redundancy and alternative energy technologies can pile considerations on top of a company exploring their data center options. Here's a short guide that can shed some light on the basic structure of data center power to help organizations make the right decision for their needs.
What Are Dual Power Feeds? Which One Does My Company Need?
Redundancy is crucial selling point for today's data centers. The business-critical information and infrastructure support systems in these facilities are simply too important to depend on the fragility of a single connection. It is not as if one can simply wander down to the basement and fiddle with the fuse box until the lights come back on.
Unplanned data center downtime costs an average of $7,900 per minute and $690,200 per outage incident, according to the Ponemon Institute. During this time, a company could take hits to its reputation and information security, as well as its bottom line. Failsafe power is critical.
Many data center providers offer a ""dual feed"" as one of its services to convey that they offer redundant power. However, there are a variety of configurations that technically qualify as dual feeds, and the level of their actual independence from one another, and therefore their redundancy, can vary widely.
Some configurations can be significantly more vulnerable to certain threats, according to Data Foundry. There are a variety of components involved in transmitting power from the electric utility substation to the data center, including switches, transformers and uninterruptible power supplies.
Single Substation, Single Feed
Despite what data center providers may advertise, single feed data centers actually represent the majority of today's facilities, according to Data Foundry. In this configuration, a single substation from the electric utility feeds power to several transformers, which provide energy to the data center.This arrangement may appear redundant, given the presence of several transformers ferrying energy into the data center, but in reality there are multiple points of failure. With only one substation and one feed, any compromise or complication to the utility or the power line risks throwing the data center offline for a prolonged period.
Additionally, getting systems back online usually requires single feed data centers to wait on the utility. In an environment susceptible to power line-snapping natural disasters, the wait could be an extended one.
Overall, this option offers low costs and high risks. A company that doesn't depend wholly on its data center may find that the promise of low expenses outweighs the heightened vulnerability. However, this choice leaves a lot to chance.
Single Substation, Dual Feeds
The first entrant in the dual feeds service offering provides some additional failover mechanisms, but is still subject to the same feed- and substation-based points of failure as the singular substation/feed arrangement, according to Data Foundry.In this arrangement, power generates from the single substation into two different feeds. These feeds offer redundancy at the next stage, an ""MTO"" or ""ATO"" switch, which offers automated or manual change to the secondary feed if power is cut off from the primary one. This provides some measure of backup should the first feed fail. However, from the MTO or ATO, the power trip essentially functions as it did in the single/single configuration, traveling through multiple utility transformers to the facility.
While this arrangement appears to offer more backup, it is still susceptible to failure at the substation, ATO/MTO and feed levels. If the electric utility is knocked offline, the redundant feeds coming from it won't help keep the data center up and running. This is another low cost, high risk option.
Dual Feeds, Dual Substations
The dual feed, dual substation configuration alleviates one of the major pain points in the previous two entries - reliance on a single substation. This can effectively put data centers and their tenants at the mercy of electric utilities. Many people living and working on the eastern seaboard of the U.S. during the aftermath of Hurricane Sandy know all too well how long a power outage can last.The model of a primary and secondary feed converging at the ATO/MTO is similar to the single substation/dual feed model, except that each substation generates its own, according to Data Foundry. If one fails, the other can be automatically or manually configured to run, providing uninterruptible, redundant power.
However, the ATO/MTO and the single feed that travels from it to the data center transformers both represent single points of failure. If anything should happen to either one of these system components, data center power would have to shift over to a generator for an unknown amount of time. This option offers additional redundant components at the key area of the utility, but there are still points of failure on the power supply chain. It is less expensive than the fully redundant model, but leaves data center operators and clients vulnerable to several single points of failure.
Dual Substations, Dual Feeds
While the other dual power feed models may purport to offer full redundancy, this is the only configuration that truly offers resiliency at every step of the power supply system. In this system, there are two substations, eliminating downtime resulting from a problem at one of the electric utilities, according to Data Foundry.This model also does away with the ATO/MTO switch, as there is no longer a need to switch over from the primary feed to the secondary one. Instead, each substation's feed provides power to each data center transformer, located inside the facility and controlled by a facility operative or a sophisticated management system.
This option offers the most control and depends less on utility companies for continued uptime. The lack of single points of failure can significantly reduce the likelihood of prolonged downtime from an unexpected outage. This option is also the priciest, and is generally restricted to state-of-the-art facilities supporting companies in industries like banking and finance, which could have significant global consequences if pushed offline for an extended period.
Clearly, not every electrical configuration is created equal, and it is important for companies to do their research so that they select a facility that corresponds to their needs. Data center power will likely continue to be the subject of much discussion, especially as alternative forms of energy, including biomass, wind and solar power, continue to be part of data centers in experimental forms. Hopefully, this guide can help break down some of the complexities of data center power arrangements."
