Connections +

When the Lights Go Out

All the network engineering undertaken to ensure network performance, quality of service, and availability will be for naught without power. Proper steps must be taken.

November 1, 2003  

Print this page

Electricity holds a unique place in our infrastructure and consciousness. On the one hand, we rely completely on electric power to run the very core functions of civilized society. On the other, we have taken it for granted.

Or at least we did. Recent events have shaken that sense of complacence. From the “brown-outs” that rolled across California in summer of 2001, to the massive outages that darkened Ontario and the northeastern U.S. on August 14, 2003, it’s clear that as much as we rely on the public utility, it is vulnerable.

Pedestrian though electric power may seem, it is the first domino of critical infrastructures. The August blackout shut down more than 100 power plants in the U.S. and Canada, knocked out power to 50 million people, and sparked an 0.7-per cent drop in economic activity in Ontario, the biggest monthly decline in more than a decade.

Although some analysts felt the Canadian economy could absorb the anticipated multi-million dollar losses, Anderson Economic Group considered all possible impacts and forecast a potential impact closer to US$5 billion for Canada alone.

As our society marches deeper into the digitally dependent age, we have increased our reliance on electric power at the same time that our demands are undermining the ability of the power grid to supply it.

While government and public utility companies need to work on strengthening the grid, private sectors has to take the responsibility of guaranteeing supplies of critical power to their own operations.

Just how prevalent is the risk?

A telecom node that handles more than a gigabyte per second (Gbps) of data for much of the working day, or a megabyte per second (Mbps) all day every day, or more than 10 kW of broadcast signal over the airwaves, probably plays an important role in the daily lives of many people.

A typical telephone exchange requires about 100 kW of power; a television station, about 100 to 250 kW; a radio station, 20 to 100 kW; a cell tower 10 to 20 kW. In data centers and wireless networks especially, the number of “critical” nodes is mushrooming.

Public telephone and data networks have long been deemed “critical” and granted some degree of power protection.

Sketchy Standards

Yet even for “critical” telecom services, there are few standards that specify minimum requirements for backup power, and existing standards are generally sketchy. For instance, the North American Emergency Number Association (NENA), which represents 911 call centers, suggests “a minimum of 15 minutes of emergency power” and acknowledges that, budget permitting, an hour would be dandy.

Without further specificity, members are encouraged to plan for more prolonged power outages and to equip call centers with a source for long-term emergency power, either a redundant utility power feed or generator. Such recommendations are a beginning, but hardly a rigorous standard.

The situation is even more tenuous in private networks. Not bound by the governmental scrutiny and up-time standards that apply to public networks–private networks are often surprisingly under-protected, relative to their importance.

Tens of thousands of private nodes route data volumes comparable to public network nodes–or greater. This represents a new “critical” class that has received far too little notice in power planning.

There are reasons why generators and surge suppressors are not enough. Utility voltage can fluctuate five to eight per cent from absolute specification. That range may exceed manufacturer specifications for sensitive electronic components, which are getting more finicky with each advance in miniaturization and processing power.

“Even though vendor specifications may say that a particular device behaves properly between 100-130V, for example, in my empirical experience, the equipment starts having trouble whenever you drop out of the normal 110-128 range,” says Jim McQuire of The Power Place, an Atlanta-based power quality audit and consulting group.

“The farther you stray from that normal range, the more problems you have. If you have 100V coming in, the equipment is not going to perform up to specifications.”

Input power is routinely plagued with power surges, sags, electrical noise, harmonics, load fluctuations, and other interferences.

A commercial customer on typical utility power will be subjected to these power anomalies daily and four to 15 complete outages per year.

Most network providers have backup generators that provide emergency power within 10-30 seconds and surge suppressors that absorb potentially harmful electrical spikes, such as from lightning.

However, these are band-aid solutions for systemic problems.

Backup generators address the most obvious power problem — complete loss of utility power — but provide no protection against the other power disturbances. It’s not enough to switch to backup power in 10 seconds, when that interval is enough to lock up a server that runs critical call-processing functions, or to lock up transport systems and fail to deliver up to SLAs (service level agreements).

Surge suppressors address the power surges, while having no effect on the under-voltage and variance conditions that can erode equipment health over time, or zap it in an instant.

Uninterruptible Power Systems (UPSs) augment and supersede these power-protection strategies. UPSs can protect equipment from the full range of potential power anomalies — not just outages, surges and spikes.

The term UPS typically refers to an AC device, used mainly to power computers and some enterprise networking devices. DC power sources perform a similar function in the telecom world.

At its most basic level, a UPS performs two primary and complementary functions:

“Conditioning” incoming power to smooth out the sags and spikes that are all too common on the grid, from generators, and other primary sources of power and providing “ride-through” power to cover for sags or short-term outages (say, 30 minutes to an hour), by dynamically selecting and drawing power from the grid, batteries, backup generators, and other available sources

Three key types of UPSs are in use today:

Standby UPSs are an economical solution for applications that need only minimal power protection. With a standby UPS, the protected equipment runs off normal utility power until the UPS detects a problem, at which point the UPS switches to battery power. Standby UPSs are best suited for home and small office computers.

Line-interactive UPSs regulate voltage by boosting input utility voltage up or moderating it down as necessary before allowing it to pass to the protected equipment. They are often used with enterprise network devices, such as hubs and routers, small communications systems, servers, and small workstation environments.

Online UPSs continuously condition incoming power to deliver clean, perfect sine-wave power for protected equipment. Online UPSs completely isolate equipment from raw utility power and all its irregularities–and therefore represent the only real choice for critical equipment that is sensitive to power fluctuations.

Today, some 95 per cent of UPS dollars are invested in online systems –a figure that has risen more than 15 points since 1998 and is not expected to abate.

Within any one of these types, systems can be configured for a broad range of output capacities, and multiple units can be deployed to accommodate loads up to megawatts. In this modular architecture, you can add or remove components as needed.

In fact, a redundant system can be designed to permit individual modules to be taken off-line for maintenance without removing the load from conditioned power.

Power management software can orchestrate the graceful shutdown of critical systems when power outages extend beyond the limits of backup systems.

More sophisticated systems can direct a selective, sequential shut-down of loads, so that the most critical functions receive the best protection.

Power management systems continuously monitor and diagnose the state of the utility power, batteri
es, and power sources, together with the condition of the UPS’s internal electronics.

Many organizations that rely on electronic systems are not fully aware of the potential for power problems until they happen, but by then it’s too late.

Are power specifications required by your electronic systems being met? Is the internal power infrastructure delivering up to specs? Are grounding issues causing hidden problems?

A power quality audit will answer those questions, and more–identifying problems that could undermine critical electronic systems and providing clear guidance on appropriate power protection strategies for your needs.

From the local loop to long-haul, from LAN to WAN, from wireline to WiFi… whatever the infrastructure in question, proactive planning can prevent the potentially devastating consequences of power disturbances.

Rob Woolner is the vice president of Canadian operations at Powerware, a provider of AC and DC power protection systems, power management software, and professional services, headquartered in Raleigh, N.C.