One of the biggest issues facing IT professionals is power management. Today's data centers, deployed solutions, and explosive growth in technology warrants a fresh look at the power strategies you have in place. Here are some things to keep in mind.Note: This information is also available as a PDF download.
#1: Consider managed power devices that provide granular information on device load usage
Many of today's power devices include managed versions to give you maximum control over how your power is allocated, used, and changed. Power distribution units (PDUs) in particular have advanced in this space well.
PDUs in the data center historically were simplistic in functionality and all driven by hard switches. Today, many PDUs provide individual port control, amp draw metering, grouping across devices, power event logging, SNMP management, and other features. This high inventory of features gives you enhanced awareness to the details of your powered devices, from the perspective of the PDU. Many management packs for computers can give you basic information on power events, but not on the holistic side that can be provided from the PDU.
#2: Use voltage line conditioning in remote or harsh environments
For the one-off systems, as well as any MDF cabinet or other technology deployed outside the datacenter, using a UPS (uninterruptible power supply) may not be enough protection. Most UPS units don't provide input regulation for the power being fed to the device. What this means is they have an acceptable range of input voltage, which is relayed right out to the UPS back end devices (computers, servers, monitors, etc.). If the voltage is outside the acceptable range of input voltage, the UPS device will run from battery and charge it if possible.
Different UPS units may handle this differently, but generally this is the case. When shopping for a UPS, some are available with voltage regulation (also called line conditioning) within the unit. But given your interface requirements input and output, voltage, amperage, and brand loyalty, you may not be able to get all features in the same device.
Line conditioning devices are not expensive (Tripp-Lite has a good offering, including rack-mount units) and add a great level of power protection before the UPS and other expensive IT equipment. You can also use a line conditioner to protect devices that are not on UPS protection. Consider a remote cabinet that has a collection of networking equipment; you could also have that cabinet air conditioned. The air conditioner would provide excessive drain on the battery, but you could put the air conditioner on the line conditioner to provide it clean voltage and extend the life of the air conditioner unit by not having mixed voltage enter the device.
Having deployed many technology installations internationally, I can tell you that the power may not be exactly as advertised. In a normal situation, the power would run "hot" at 136 volts AC (VAC), causing my UPS unit to run in high voltage condition. This leads to undue wear on the battery. Installing a line conditioner that accepts voltage in the range of 74 to 149 VAC would provide a smooth output much closer to the 120 VAC that the device required.
#3: Correctly provision power installations with facilities personnel
When planning an installation with the facilities staff or electrician, this is definitely the time to pull out all the stops in allocating adequate power resources for your technology needs. If the electrical staff is performing the installation, it scales much easier to add lines rather than make additional trips or add circuits after the fact. A good rule of thumb is to seek double what you currently need of each interface type. Should someone point out that you don't need that many power resources currently, point out future trends from management that may correlate to additional IT needs, as well as any migration strategies that may have a parallel old and new system potentially requiring double the power as the base needs.
#4: Consider redundancy in power sources
Take the redundancy as far down as you can, if possible. The easy strategy is to have dual power supplies on a server -- but take it down farther if you can. For example, the server with dual power supplies can have each power supply go to a different PDU, then each PDU can be fed from a unique circuit back to the main source or to different sources if two sources are available.
This can also apply to smaller rack environments. If you have multiple UPS units in a rack, but a single main source for the rack or room, make the dual power supply devices connect to two different UPS units in the event of UPS failure. This also makes UPS or PDU maintenance much easier if all critical systems are split between two sources.
#5: Establish UPS battery replacement schedules
Any battery's performance over time will decline. If you have UPS units in use in your technology environment, the batteries will eventually fail. Having a set replacement schedule for the battery unit or for the entire UPS device will provide you with the following benefits:
- You will be aware of your power situation by re-evaluating it during a battery replacement.
- You will be familiar with access procedures to your devices in the event of a failure.
- You will know that your power protection strategy works because you are diligent in administering it.
#6: Ensure strain relief is used for connections
On a power source, strain relief can be sophisticated, with access and management built into a PDU or UPS. But it can also be unsophisticated, using cable ties or fabric attraction strips. The same goes for device connections. Some servers and computers have built-in strain relief and some do not. Both the high-tech and low-tech mechanisms can be effective in protecting your devices from an inadvertent pull-out. This can happen during a rack device extension, addition of a new device, cable management practices, or simple access.
#7: Turn off retired or unused devices
This will reduce your power consumption -- and possibly accelerate your removal of the device so as not to overprovision power unnecessarily. Consider a server that is powered on, but with its network connection unplugged as a fallback just in case the new server fails. This practice is wasteful in power consumption. And should the device ever be used in a live environment, it has unnecessary runtime on the whole system.
The same goes for monitors in racks. Once systems are installed and running properly, evaluate the need for a monitor in the rack. Many systems can run headless and have a monitor connected only if necessary. Technologies such as IP-based KVM, HP iLO, Dell DRAC, and others allow systems to be fully managed without a keyboard, monitor, or mouse attached. This could re-assess the need for a KVM device in a rack for ongoing use for certain applications.
#8: Take the time to learn interface types
As your technology needs grow, you may see that your inventory of interfaces increases. When selecting devices such as UPS units, PDUs, and servers, ensure that your interface requirements are met end to end. For new installations, read up on technical requirements to make sure that the correct interfaces, voltage, and amperage are available for the installation. Always check with your facility manager or electrician to ensure that your interfaces will be available as well.
To the IT professional's eye, the L6-20P and L6-30P interfaces look similar. Usually, cable is thicker on the L6-30P. But you can have a project-stopping surprise if you are planning to install a device requiring a L6-30P source when a L6-20P is in use.
#9: Sample voltage regularly
Ensure that your input voltage is maintained as you are expecting. Any drop or rise in the input voltage that is out of the acceptable input ranges will cause undue wear and tear on your UPS device or line conditioner. Raise any concerns to your facility manager or electrician to ensure that there are not any larger concerns from usage or growth over time.
#10: Identify the most critical devices and fully protect them
Put your "gold" devices -- those that require the most runtime and the most power protection -- in situations where you will have all resources available. For example, if you can not line condition and UPS protect all servers in your installations, consider putting your most critical devices on the best power equipment you can provide. This would include also the highest battery runtime, most management of UPS or PDU devices, and the newest equipment for battery condition.