A well-built power protection solution, featuring high-quality, highly efficient UPS hardware, can help keep your business applications available, your power costs manageable, and your data safe. When planning or updating your power protection solution, how do you choose the best UPS for the job?
The answer depends on a combination of factors that have been influenced by industry trends and technology advances. To ensure you always end up with the right UPS for your needs, remember these eight important considerations during your selection process:
Question number one is: Should you get a single-conversion, double-conversion or multi-mode UPS? The answer depends largely on how important energy efficiency is for your organization relative to protection.
Single-conversion UPSs are more efficient than double-conversion devices, but offer less protection. This makes them a good fit for loads with a higher tolerance for failure. More specifically, standby UPSs (the most basic type of single conversion UPS) are generally the best option for smaller applications, like desktop and point-of-sale solutions, while line-interactive UPSs are typically preferable for smaller server, storage and network applications located in facilities with access to relatively trouble-free AC utility power.
Double-conversion UPSs, which provide the highest levels of protection, are less efficient but are usually the standard choice for protecting mission-critical systems.
Multi-mode UPSs, although they may be more expensive than either single or double conversion systems, are the best choice for companies looking to achieve an optimal blend of both efficiency and protection.
When the utility generates power, it is at the three-phase level. This type of power is available to almost all commercial and industrial customers, as they are typically large consumers of power. Single-phase UPSs tend to be a sensible and economical option for simpler, smaller applications with low kVA requirements, which are typically seen in the home, small businesses and in remote or satellite offices where computing power is less than 20,000 VA. Three-phase UPSs are generally the preferred choice for high kVA applications, which are typically more sophisticated and have high compute densities. Large multi-story buildings, data centers and industrial facilities protecting high-power processes are typical three-phase UPS customers, as they need to distribute large amounts of power over relatively long distances.
A UPS’s rating is the amount of load, in volt-amperes (VA), that it’s designed to support. UPSs are available with ratings as low as 300 VA and as high as 5,000,000 VA or more. Use this very basic procedure to determine the approximate UPS rating your organization requires:
1. Make a list of all the equipment your UPS will be protecting.
2. Determine how many volts and amps every device on the list draws.
3. For each device, multiply volts by amps to arrive at a VA figure.
4. Add all of the VA figures together.
5. Multiply that sum by 1.2, to build in room for growth.
The UPS you buy should have a rating equal to or greater than the final number you arrived at in step 5, unless you have more precise load data for the equipment you are protecting. Here are a few additional considerations to keep in mind:
Relying solely on nameplate ratings may lead you to oversize the UPS system, so always use your equipment manufacturer’s sizing calculator tools as well, if available. Most major manufacturers have Web-based or downloadable sizing tools that can closely estimate your equipment’s power draw based on the configuration you are using.
When deploying centralized power protection architecture, you typically deploy larger kVA UPSs than you would deploy using a distributed power protection scheme.
If your UPS will be supporting motors, variable-speed drives or laser printers, add more VA capacity to your requirements to account for the high power inrush that occurs when those devices startup. Your UPS vendor can assist in applying the proper UPS and rating for these types of applications.
Companies that anticipate rapid near- or medium-term growth should use a multiple higher than 1.2 when building in room for growth in the procedure above. Organizations that expect to upgrade their server hardware soon should also consider this, as newer servers tend to have higher power requirements than older models.
UPSs come in a range of form factors that fit into two master categories: rack-mounted and freestanding. The largest UPSs aren’t available in rack-mounted form factors, so companies with substantial power requirements almost always use freestanding devices. For companies with more modest needs, deciding between rack-mounted and freestanding UPSs is largely a matter of data center design philosophy. Some organizations use rack-mounted UPSs in an effort to consolidate as much hardware as possible in their enclosures. Others prefer to maximize the amount of rack space available for servers by using freestanding UPSs. From a technical and financial standpoint, neither approach is inherently superior to the other.
Organizations can utilize a variety of deployment options, technologies and services to increase the reliability of their power protection solution. Here are some of the most effective ones:
Redundant deployment architectures: Deploying UPSs in redundant groups can increase availability by ensuring that critical loads remain protected even if one or more UPSs fail. There are three main kinds of redundant UPS architecture:
Zone: In a zone architecture, one or more UPSs provide dedicated support for a specific set of data center resources. That way if a UPS fails during a power outage, the impact is limited to the zone that device supports.
Serial: In a serial architecture, multiple UPSs are connected end to end such that if any one UPS in the string fails, the others can compensate automatically.
Parallel: Parallel architectures use multiple independent parallel-connected UPSs to achieve increased redundancy. If any UPS fails completely, the other systems can keep protected information technology equipment (ITE) loads operational.
Hot-swappable components: Technicians can repair or administer a UPS that uses hot-swappable components without powering the UPS down, avoiding the risk of downtime.
Extending battery runtime: A typical UPS battery provides five to 15 minutes of backup power. Organizations that need more than that can use supplemental external battery modules or cabinets to add as much as several hours of emergency runtime at full load.
Battery management: One of the most important parts of the UPS system is the energy storage system, which is typically a battery. Many UPS systems continually “trickle-charge” the battery, which has a tendency to degrade the battery’s internal chemical composition, reducing battery service life. While large banks of flooded electrolyte batteries for high power (greater than 500kVA) UPS systems need to have the battery trickle charged, most non-spillable (VRLA, see “UPS energy storage” below) batteries used in today’s lower kVA UPSs can benefit from a charging technique where the charger turns off and “rests” the battery.
Some manufacturers have labeled this charging technique as Advanced Battery Management. This technique helps increase battery service life by as much as 50%. Batteries utilizing this technology last longer and are more reliable thanks to the UPS system’s three-stage charging technique, sophisticated sensing circuitry and an automated battery test routine that notifies the end user of deteriorating battery conditions that warrant replacement.
Remote monitoring: The best way to address UPS problems is to prevent them from happening in the first place. Remote UPS monitoring applications continually watch for warning signs of future trouble, such as deteriorating performance or an overheating battery, and send real-time notification when potential issues develop. That enables technicians to make repairs before serious breakdowns have a chance to occur. Data centers can perform remote monitoring themselves or contract with an external provider to do it for them.
Implementing a robust power protection solution takes time and money. To get as much value as possible from that investment, companies should estimate their needs for the next three to five years when comparing UPS options. If you’re likely to have significantly larger power requirements in that timeframe, choose appropriately larger UPS hardware.
Data centers with fluctuating needs or future requirements that are difficult to forecast can employ two strategies to increase the scalability of their UPS deployment:
Deploy UPSs in parallel: Parallel UPS architectures boost scalability as well as redundancy. As their power needs increase, organizations can simply add more modules to existing UPS systems, rather than replace current devices with new ones.
Use modular UPS products: Some newer UPSs feature modular designs that allow you to add capacity incrementally as requirements increase. For example, some such systems provide up to 50 or 60 kW of capacity in 12 kW building blocks that fit in standard equipment racks. As requirements increase, another 12 kW unit can simply be plugged in. Even the largest UPS systems can be made modular in 200 to 300 kW increments. That’s a scalable and efficient approach to keeping up with escalating power needs that also lowers upfront capital spending and conserves data center floor space.
Even with a UPS, your IT system could still go down in the case of an extended power failure or if the UPS gets overloaded for too long. Communication software can not only provide real-time notification of UPS status, but also lets you assign automatic actions to perform in case of a power event. This is extremely useful if your system operates continuously without users being present to manually shutdown affected equipment.
Virtualization is now bringing a new set of complexities, as the bond between operating system and physical hardware is no longer the standard. Some suppliers of UPS software must ensure that shutdown software agents are installed on each virtual machine as well as on the host machine. This can be quite tedious if the number of virtual machines is large, which is becoming the standard in many virtualized environments. Leading-edge UPS manufacturers have developed new software platforms that reduce this management complexity by integrating their software into virtualization management platforms like VMware’s vCenter or Citrix XenCenter. In these environments, one single software installation can control and shutdown any cluster of servers. Another benefit is the enablement of automatic live migration of virtual machines in case of a power outage, as you are no longer limited to the option of shutting down the servers and stopping operations.
To summarize: logical and complete power management applications can help companies:
Monitor and administer their UPSs from any location with Internet access
Automatically notify key personnel of alarms or alerts
Perform orderly, unattended shutdowns of connected equipment, or better, work with virtualization software to move virtual machines so as to maximize availability of key applications and hardware
Selectively shut down non-critical systems to conserve runtime
Analyze and graph trends, to predict and prevent problems before they happen
Integrate with existing network and management systems via open standards and platforms
With proper servicing, a well-made UPS can operate safely and reliably for as long as 20 years. Without proper servicing, even the best UPS is significantly more likely to fail when you can least afford it. Companies in the market for UPS hardware, therefore, should also choose an appropriate UPS service plan from a service provider with the experience, know-how and resources to provide comprehensive, high-quality support.
So, which UPS topology is right for your data center? Where there once was only one “right” answer, new technologies offer effective new choices specifically designed for high-efficiency, high-density data centers.
With best practices and the right choices in equipment, data center managers can reduce energy consumption by nearly 50%. That means almost three-quarters of the power utility bill will fuel actual IT processing, compared to less than 50% of the power supplied to a normal data center today. With a more efficient allocation of power, you not only reduce utility bills and total operating cost, but also achieve more with available backup power and cooling systems— ensuring the efficient, reliable operation of your data center well into the future.
Ed Spears is Product Manager, Power Quality Solutions Operation, Eaton Corporation