As businesses become increasingly dependent on electrical and low-voltage systems for both building and business operations, it becomes more important that those systems remain operational and accessible at all times. This need is at the top of the list of facility concerns because downtime can cost a lot of money. Dan Carnovale, power quality solutions manager at Eaton Corp., Moon Township, Pa., explained that reliability is of the utmost importance and an overt issue within data centers and other at-risk industries.
“From the data-center perspective, and also within industrial environments, there is a high emphasis and cost benefit of reliability in terms of redundancy,” Carnovale said.
Because of the inherent design of data centers and the baseline functions of their operations, redundancy is essential. They have become so integrated that, in some cases, servers facilitate, essentially, all operations. If those servers go down, a company could be crippled.
When it comes to data centers, redundancy basically means a data center has at least two sets of every component, including uninterruptible power supplies (UPSs) and power distribution units (PDUs).
“In a typical data center, you have servers with two power cords leading to two independent power supplies, each separately supplied by alternate sources, often called an ‘A’ and a ‘B’ source, then two PDUs feeding those sources with a UPS system on each side,” Carnovale said. “Finally, the utility service may also be redundant and is likely backed up by multiple generators. You get redundancy all the way down through the system.”
The requirement for such redundant systems is based on the utility source that feeds the data center.
“Standard utility service just cannot be relied upon alone, and just a single source of redundancy is not enough,” he said. “It all comes down to technology and redundancy versus cost. You need to gauge the availability of what can be considered as acceptable downtime. Within a data center and other industries, that acceptability becomes obvious, based on operations and the cost of downtime.”
Regardless of their size, location or physical layout, the type of business or entity using the data center determines acceptable levels of redundancy and reliability. Carnovale explained how different vertical markets mandate different redundancy requirements.
“Take the petrochemical industry, for an example,” he said. “All processes are critical. Safety issues are often the main concern if the power goes out, and therefore, any outage is unacceptable. They must have the ability to run 24/7/365.
“The government is another entity that is concerned with downtime but not because it is costly, per se, as in other industries, but rather because disruption may cause failure of their ‘critical mission.’ Take the military operations and, say, government hospitals. Those are examples where it is not simply about costs but rather the fact that they absolutely must be always running, have redundancy to do so and be able to complete their missions at all times.”
The facility and layout of a multibuilding campus, such as a large hospital or university, also plays an important role in what is optimal redundancy to support all systems.
“Within both the healthcare and the education market, you have a lot of aging infrastructures out there,” Carnovale said. “You can use redundancy to bring those facilities to a higher level of reliability. For instance, take a 50-year-old dorm. It is not equipped for the 18 or so electronic appliances that each college student brings with them, on average, to school nor are the classroom buildings able to handle the computer and research labs filled with similar electronic loads. The infrastructure does not support true reliability. Because of this, you add in backup systems, generators and various other redundancy measures to make everything—all systems—more reliable.
“Retail is another market where redundancy is considered, but often the cost benefit is not there. In retail, equipment and systems maybe get replaced every 3–5 years, and retailers are not going to pay to invest in significant redundancy. It simply is not a good return on investment. Small UPS units are typically the extent of redundancy in retail,” he said.
Although every facility using a data center for computing, information and record storage assumes that its own operations are of the utmost importance, some places mandate constant reliability. For instance, when involving financial or health issues, the power must stay on.
“It is critical that power within a facility does not go down, especially in a data center or where things like financial records are kept. Places like banks and trading floors cannot go without power and go down,” said Bill Reid, marketing manager, Siemens Industry Inc., Building Technologies Division, Buffalo Grove, Ill.
Reid agreed with Carnovale that the issue of redundancy is imperative based on the type of facility. Properly placed redundant systems can make the difference between meeting quotas or keeping business running, and it can also make the difference between life and death.
“Take hospitals and operating rooms. What happens when the power goes out? How about industrial plants? Steel mills, they lose power and the steel begins to harden and can no longer be used,” he said.
Reid said the healthcare industry is the most critical venue for reliable redundancy, when compared to other noncritical operating environments.
“Most people think that data centers require the most reliable power,” Reid said. “But, when you look toward the future, you see the most demand coming from the healthcare industry, and a lot of that has to do with how healthcare is moving toward being more digital in nature.”
The advent and widespread adoption of electronic medical records, digital imaging and computer-based medical systems means that healthcare facilities not only have continual, reliable power at all times, but that redundancy has been placed, in effect, for all systems. Both integrated and stand-alone systems need redundant systems for the highly computerized and digital medical facility to maintain operations at all times, regardless of extenuating circumstances or external conditions.
The trend of constructing extremely large data center facilities started in the dot-com era. Initially, the goal was to ensure continual operations at essentially any cost, which resulted in vast amounts of redundant systems being installed in the massive computer centers that acted as the brains and hearts of all operations.
“What you have now are data centers that run off of 500–600 watts [W] per square foot,” Reid said. “The typical commercial building these days operates at around 30–40W. That is a huge difference.”
As time passed and the economy changed, data center operators began to realize that, even though redundancy and reliability remained vital, such measures could be scaled back with acceptable results. That large-scale power demand has been cause enough for data center designers to rethink the term “necessary,” and many systems, such as redundant ones, are being re-evaluated based on their associated cost.
In data centers, cooling is one of the primary power costs because the equipment contained within both gives off and is susceptible to heat. Cooling costs have increased because the servers have become more dense and higher powered (thus, they operate at a higher heat level compared to their nonblade counterparts). As a result, the servers generate massive amounts of heat and require more cooling power to keep the data center facility at an optimal temperature for operations.
The cooling systems within a data center also must remain operational and, therefore, must be redundant. The heat generated throughout the center would be detrimental to the equipment if the cooling system were to go down and remain offline for any length of time.
A continual evolution
A final piece of this puzzle is a change in the way designers and integrated systems contractors view the need for complete and total redundancy. Now, the school of thought is that redundancy is not needed on all systems.
Data centers are categorized by a four-tier system, which encompasses the redundant power systems. Tier 1 and 2 involve single paths for both power and cooling redundancy, while tier 3 and 4 data centers operate off of multiple paths.
“Initially, tier 3 and 4 data centers had multiple redundancies for all systems,” Reid said. “Everything had a backup source, and especially at the tier 4 facilities, you would have this warm and fuzzy feeling that nothing would ever go down.”
Reid explained that new data centers are now being downgraded to more of a tier 1 or 2 level, since the majority of systems that go down only cause inconvenience and not a financial meltdown or other negative effect.
“Places like national defense sites that need to be able to operate even if a bomb goes off have to remain at that tier 4 level, but other places can get by going down to tier 1 or 2,” Reid said. “It is just cheaper, and for the inconvenience, the cost does become a factor.”
“In terms of redundancy, what you see happening in data centers are what we call home-grown solutions,” Carnovale said. “What they do is use a little of this, a little of that, and they create their own redundant environments. One reason you see this happening more and more is that data centers are competitive, and they just do not want one another to know what they are doing in there and how they are keeping their systems operational.
“The loads placed on data centers these days are much more taxing than in the past. You have a lot of electronics in there now, and those bring with them a higher susceptibility to power issues.”
It is those power issues that then cause outages, and redundancy becomes an even more critical issue, as it staves off such power disruptions that then result in system downtime.
However, while the purpose of limiting downtime is to control costs, building redundancy into any system is costly. Therefore, riding on the theme of the times, redundant design has embraced efficiency and as a result, facilities have adopted hybrid solutions.
“Today’s data centers are 20–30 megawatts, the size of yesterday’s industrial plants,” he said. “Based on their size, even the most highly reliable data center may want to save energy as well. So with hybrid solutions, they can do that. For example, you can have a large UPS that runs in a high efficiency mode and provides 99 percent efficiency, or you can run the same unit in full online mode at 94 percent or use a combination of both methods. For many, a few points of efficiency can save hundreds of thousands of dollars per year and offers acceptable redundancy. A lot of today’s solutions are tied into the building management systems. The ability is there to monitor power flow and reliability. You have the ability to look at and monitor power quality, lighting, HVAC, alarms and controls. That tie-in, and integration then allows alerts to be sent multiple places at the same time to get problems solved.”
Quick service redundancy
One way in which redundancy can be implemented at a lower cost is through the use of modular-style data centers and computer rooms. As customers grow their computing, data storage and networking needs, contractors can and will find themselves becoming increasingly involved in this area, Modular data centers are basically built inside shells that are moved on-site to customers, thus providing them with a self-contained, prebuilt data center that can reside right outside a facility. This data center technically exists off-site; however, no additional space, facility, land or property is required. The units are brought in on tractor trailers and are basically plugged in to a main power feed, networked in to the main building, and are up and running within hours.
Reid said that these new versions of data centers, even though much smaller than free-standing, larger data centers, can be useful when adding redundancy to a facility at a lower cost.
“These things cost $5 million, not $200 million for a standing data center,” he said. “Now, they are not tier 4 level centers, but they have the ability to solve the backup problem. They then can be expanded or even swapped out as needed and, thus, provide quick service redundancy as needed.”
Electrical contractors have the power
Contractors can find various avenues to pursue in terms of helping their customers achieve and maintain redundancy rates and solutions that suit their individual needs. Carnovale emphasized the important role contractors play in making sure facilities are built with reliability and redundancy in mind from the beginning.
“The entire system needs to be laid out, designed and implemented properly up front, and it needs to be commissioned,” he said. “When you hear of issues, it is usually a system where a bunch of parts are just put together without regard to how they will interact with each other. What happens is that you sacrifice reliability when systems are not implemented well.”
Redundancy is only effective when installed and maintained properly, and it’s a service that savvy electrical contractors can offer and generate profits from.
“You need to constantly monitor those systems,” he said. “Recommissioning is probably the biggest opportunity as frequent checks and rechecks ensure that the systems are operating as designed, reliably and efficiently. In fact, there is a huge opportunity there for contractors. You need to track and monitor on an ongoing basis. Look at the statistics. The Electric Power Research Institute says that, in a typical year, $119 billion is lost in business revenue just based on disruptions. It has a direct impact on business.”
STONG, a freelance writer, lives in central Pennsylvania. She can be reached at firstname.lastname@example.org.