Home > Cover Story
 Print Friendly Page ||  Email this story

Cover Story: Power Conditioning
Powering up the business

Power conditioning is essential for any enterprise that wants to provide five-nines uptime. Here are some typical power-related problems and ideal strategies to counter them. by Soutiman Das Gupta

The word 'outage' is likely to scare a CTO more than the man-eating shark did in the film Jaws. And why not? An outage in your IT infrastructure will result in loss of productivity, management dissatisfaction, loss of revenue, loss of transactions, and may lead to customer dissatisfaction.

The shocking truth

System downtime may be the result of various factors: the high-tech ones like network failures or hardware/software crashes, or the natural ones like earthquakes, floods, and fires.

But according to the "Cost of Downtime" survey conducted by Contingency Planning Research Inc., power-related problems were the most frequent reason for outage in an enterprise.

The survey also showed that, power outages took place 25 percent of the time, and power surges/spikes three percent of the time. Together (28 percent), they comprised the most common outages, and were followed by storms, floods, hardware errors, and fire. Network outages occurred only two percent of the time.

This shows that while CIOs/CTOs have been busy creating redundant network architectures with no single point of failure, the aspect of power supply and solutions took a 'bit of a back seat'.

Defining power conditioning

Power conditioning is a well-defined strategy to ensure that an enterprise has continuous power availability, which is clean, steady, and free of irregularities. It also encompasses power redundancy in terms of backup and alternate supply.

Enterprise power conditioning is a subset of Business Continuity Management (BCM). BCM's goal is to provide 100 percent business uptime, of which power conditioning is an important aspect. This is because electricity is the basic necessity to run the hardware and the applications, which enable a business.

Another survey jointly conducted by MAIT, Emerson Network Power (ENP) India, and Feedback Consulting in 325 Indian enterprises showed that over 60 percent firms experienced power disruption more than once a month. And India Inc. could be losing over Rs 20,000 crore annually, in direct losses, due to poor power quality and operating environment related downtime. The final bill for India Inc. would be much higher if the indirect losses are taken into account.

Beyond an UPS

An UPS (Uninterrupted Power Supply) is the core component of a power conditioning strategy. It has features like the inverter, voltage stabilizer, and EMI/RFI filters. It also has a number of built-in intelligent features, which take care of most of the typical power-related problems (See box: What can go wrong with power?). UPSs can also be managed remotely through browsers. And this provides a lot of flexibility.

But one has to look beyond a UPS in order to deploy a total power conditioning strategy. An enterprise also needs to plan redundancy and backup at various levels of operation. Redundancy should also be built into each zone and into all pieces of equipment. Ground faults and wrong wiring issues have to be dealt with. And the success of the solution has to be reviewed through audits and checks.

Besides, older UPSs do not usually take care of all the power-related problems. For example, a UPS must have an isolation transformer (to protect from EMI/RFI noise) in its output circuit to qualify as a power conditioner.

Click on image for larger view
An ideal power conditioning strategy

An ideal power conditioning strategy

An ideal strategy is one that encompasses all power-related problems. In large enterprises, power conditioning is usually the shared responsibility of the Operations Manager and the IT Head. In smaller enterprises, the IT Head should be the one to swing into action.

It is difficult to suggest an ideal strategy since different companies from different verticals have different business and operational
needs. But one may use a broad framework.

1. At the point/gateway where the power enters your enterprise from the electric supply, there needs to be an automatic transfer switch. This is because most companies will have a backup diesel generator set. The transfer switch swaps between the two feeds. You can install a manual transfer switch, but it needs a person to be deployed at the site around the clock. An automatic transfer switch is useful especially in the case of remote locations.

2. Power from the transfer switch flows into a surge suppresser. This controls any high power fluctuations, which are likely to damage equipment. It has somewhat the same functions as that of a domestic PC spike buster, but on a larger scale.

3. AC power now passes through an UPS which has a battery backup and automatically switches over to the alternative supply in case of outages. The power is now distributed to various departments and sections of an enterprise through a power distribution cabinet. Some telecom switches and equipment require DC supply. In this case the company needs to set up DC power systems and interfaces.

4. Cables must be robust and the conduits and pipes must be laid according to safety principles. The embedded AC/DC power supply is also critical. This is the power supply grid present inside servers, switches, and other devices. Critical hardware should have dual power grids, so that one acts as failover.

5. The customer also has to evaluate and identify the critical areas for which uptime needs to be enhanced. There may be possibility of distribution faults or some fault in the facility. An enterprise can deploy dual power supplies, dual distribution equipment, and static switches at the load end.

6. The capability to monitor operations from remote locations has emerged as an important feature for any solution. So, all these solutions should allow browser-based monitoring. Information of impending failures like a weak battery bank and alarm conditions which need manual intervention, can be retrieved from anywhere in the world.

Some management software can also send SMS and e-mail messages as alerts. The remote monitoring and message delivery functions should be closely integrated with the customer's backend network.

7. As a part of a complete power solution offering, all the equipment has to be wrapped around by services. Servicing starts right from pre-sales and carries on as a life-long commitment made by the power vendor to its customers.

Assessing downtime costs

Manoj John, Industry Manager, Industrial Technologies Practice, Frost & Sullivan (India) feels that the extent of damage to an organization in case of a power outage differs from one company to the other. Downtime cost can be calculated using a formula based on a number of common cost factors which reflect business realities.

Some common cost factors are:

  • User productivity
  • Revenue or transaction impact
  • IT productivity
  • Lost future revenue
  • Market impact
  • Fees/penalties/other charges

In functions like ATM (Automated Teller Machine) transactions, mobile telecom, rail reservations, and air reservations, downtime has direct impact on revenue and goodwill. The cost can be calculated by multiplying transaction volume per minute with the number of minutes lost in downtime. This revenue is lost forever.

The MAIT and ENP survey reported that over 50 percent of the respondents mentioned impact on business processes, PCs, and servers as the most severe manifestation of downtime. In today's networked environment, any break in business continuity not only results in monetary loss, but also erodes customer confidence and adversely impacts the image of the organization.

Soutiman Das Gupta can be reached at soutimand@networkmagazineindia.com

What can go wrong with power?

Anomaly in electrical supply is usually unnoticed, perhaps because electricity works silently and invisibly. But a lot can go wrong with it. Here are some typical power-related problems.

Surge: A short-term increase in voltage which lasts at least 1/120 of a second. Surges result from presence of high-powered electrical motors like air conditioners and electric pumps. When this equipment is switched off, the extra voltage is dissipated through the power line.
Spike: A very high but momentary rise in voltage. The worst spikes are from lightning strikes on the power wiring which can damage your servers seriously. Spikes can also originate from the power grid. Spikes and surges can progressively damage power supply and other components. So, after a number of spikes and surges, your server may just die. You may lose a power supply, your motherboard; and even your hard drive and everything on it. Fortunately, direct lightning strikes to power lines are rare because a power line is usually well isolated from earth, and lightning looks for an earth.
Over-voltage: Sustained voltages that are higher than normal. Over-voltage causes a computer's power supply components to overheat and die.
Brownouts: Also called 'sags', a brownout is a sustained decrease in voltage level. If the voltage is too low for the computer's power supply to compensate, the computer can freeze or behave erratically. Erratic behavior is the worst because it can cause major data corruption on your hard disk. Hard disk drive motors may also overheat. Over-voltage is the most common power problem. It's usually caused by the start-up power demands of electrical devices like motors, compressors, elevators, and shop tools. The closer to maximum capacity a power supply is, the less likely it is to handle a given surge or sag. A computer with a 300 watt Power Supply Unit (PSU) is likely to deal better with line irregularities than one with a 235 watt PSU, although it may not ever need more than 200 Watt of the PSU's possible output.
Power failure: Any loss of power of more than 1/120th second. In other words, the plain old power blackout. Servers may lose data in the RAM, cache, and FAT (File Allocation Table).
EMI/RFI: Electro-magnetic Interference/Radio Frequency Interference may occur due to electromagnetic noise from devices like printers, radio transmitters, and industrial equipment. It may cause mysterious hangs and other problems.
Switching Transient: Instantaneous ender voltage (tapering off), which is shorter than a spike and may occur for a nanosecond. It can result in quirky computer behavior and puts stress on components, which can lead to premature failure.
Harmonic Distortion: Distortion of the normal power waveform. This happens due to the use of variable speed motors, disk drives, copiers, and fax machines. It can cause communication errors, overheating, and hardware damage.
Wrong Wiring: Wiring may be deployed incorrectly, but the equipment may work when plugged in. The equipment will face the risk of shock and early failure.

Ground Faults: Faulty earthing of equipment. This can cause mysterious malfunctions and destroy network equipment seemingly, without explanation.

UPS topologies

Manoj John, Industry Manager, Industrial Technologies Practice, Frost & Sullivan (India) says that UPSs can be deployed in three topologies: online, offline and interactive.

Online UPS protection provides the highest level of power quality protection, power conditioning, and power availability. In an online UPS, the inverter supplies conditioned AC power to critical equipment even when the mains supply is not available. And raw power from the main source is used by a rectifier-cum-battery charger to power the inverter. The load equipment will never receive raw power from mains in any condition.

Offline UPS protection, also called standby, is a cheap and cost-effective choice for small, non-critical and stand-alone applications. In this configuration, mains raw power is continuously supplied to the load till it is available. The inverter is normally off. The inverter will start only after the mains power fails and there is a relay changeover with a small break in output power. The break in power is normally shorter than that required to stop or reboot the computer operation.

Line-interactive UPS protection provides effective power conditioning, which is better than offline or standby UPS and backup. This is particularly suitable in areas where power outages are rare, but there are wider and frequent voltage fluctuations. In this configuration also, power is normally fed by mains but through a voltage stabilizer. The inverter runs and supplies power only during mains failure conditions.

Conditioned environment

Servers and other hardware also need environment conditioning to perform optimally. S.S. Bapat, Country Champion, Uptime Solution, Emerson Network Power (India) Private Limited says, "Today's servers and communication switches generate up to ten times the heat per square foot as systems manufactured just ten years ago and that is dictating new approaches to heat removal."

For example, batteries used in a UPS are one of the many pieces of equipment affected by severe heat. The optimal battery temperature is 77° F. At 111° F the design life of the battery is reduced by more than 80 percent

The study of environmental conditioning is very extensive and elaborate, and will need an entire cover story dedicated to it. However, in short, servers and other computers ideally need low temperatures, controlled humidity, and flow of air over its chassis. It's common knowledge that domestic type air conditioners can cause static electricity on the hardware because they reduce the humidity in the air. Servers and other hardware need normal humidity condition.

The best solution is to use precision cooling systems, which work around the clock and are customized for a particular enterprise. In data centers and server rooms, conditioned air should be pumped in from vents on the floor or ceiling to maintain free air flow. Handheld digital thermometers can be used near servers and racks to check local temperatures.