Battery Backups

Introduction

A battery backup, also called an uninterruptible power supply (UPS), is a hardware component that contains a large battery inside, capable of providing 120-volt utility power, to run computers, servers, and network equipment.

The primary function of a battery backups is to allow on-premises IT equipment to continue running during a brief power outage.

It is connected directly to utility power, and then the protected equipment is plugged into the UPS. During normal operations, the equipment draws its power through the UPS. If power goes out, though, the UPS has components inside that switch the power output (from the UPS to the protected equipment) away from trying to draw from utility power, to be provided by the battery. The power source inside the UPS is switched more quickly than capacitors discharge in typical IT equipment (all a matter of a few milliseconds), so the components in the computer or network switch do not notice any interruption in the electrical supply.

Battery backups should not relied on for running your IT system throughout a planned or unplanned power outage. For this, you need backup utility power, such as a generator. If you have a generator, the battery backup units can run the equipment in the time it takes for the generator to start up and take replace utility power. And, if the backup power system can't support your whole load, the generator will be used to support the most critical components, while the others will be shut down gracefully while on battery.

Without a generator (or other backup utility power), battery backups enable server systems and workstations to be shut down gracefully, while running on power supplied by the battery unit, in the event of an unplanned power outage. A graceful shut down means applications and databases are closed, operational data and database transaction logs written to non-volatile storage, open unstructured files saved, and any failover to systems that still have power can be done, before the computers and servers and network equipment shuts off. A battery backup system, incorporating software and data connections between the battery backup units and the data network, will have a system that automatically performs the graceful shutdown process.

Once power is restored, your company should then have personnel available to start up the system again and perform functions testing according to a checklist to confirm all applications and services are running as normal. A documented and tested disaster recovery plan should be available in the event of any problems restarting the systems.

Modern office buildings, especially the large towers, perform hours-long power shut-downs these days at least twice a year in order to test backup power and emergency communications systems. These tests require any company with an on-premises IT systems to perform their shut-down and recovery procedures if a generator is not available.

While not often discussed this way, a battery backup system may be an important component of IT security. At the very least, it can prevent loss of data should power shut down unexpectedly while a user has unsaved information open on a desktop computer, or a database server has information in an inconsistent state (such as in the middle of saving records to a storage device with a write cache). It can also in

You may see some whitepapers explaining that you should consider availability when planning for capacity. But, this isn't a good idea, as explained below.

Capacity Planning

To determine how much to invest in a battery backup, you must know the load drawn by your equipment, and how long the graceful shutdown process may take, and then use a UPS that will provide battery capacity to run your equipment for this time at a minimum. You should add contingency time though—for example, an extra half hour if your technicians would have to be contacted to travel to your office to perform the graceful shutdown.

You may then choose to add more capacity, so that you can avoid graceful shutdowns depending on the circumstances. For example, if you know that most power outages in your area are far less than one hour, so you might add battery capacity to extend run time for one hour beyond the time needed to perform a graceful shutdown plus contingencies. If you do this, then you can wait for one hour after a power outage before starting the graceful shutdown procedure, or even calling in your IT personnel.

For a workstation, you need enough time for users to save their files, close their programs, and shut down their computers. Users should always be instructed to log off their computers at the end of the day, so no files are left open. In the event of a power outage after hours, the UPS software will shut down the computer automatically, gracefully.

Capabilities

Battery backups units come with various capabilities. Some of the features include:

  • Line-interactive
  • Automatic scheduled battery self-test
  • Output quality (pure sine wave vs. step approximated)
  • Continuous online (fail open)
  • Remote management
  • Diagnostics and reporting
  • Battery charge management
  • Outlet groups, and sequenced power restoration
  • Environmental monitoring (temperature)
  • Load Shedding. You tell it to turn off the Outlet Group if a power failure last longer than so many seconds, or based on remaining UPS runtime, and whether to stay off or turn on when power returns.
  • Calibration

The most low-grade units may fail at any time, causing your computer to shut down. Or, even though there is no indication the battery is dead, your computer works until power goes out, in which case the UPS and your computer shut off.

Diagnostics and Testing

Setting your equipment in a consistent-state and then letting it run on battery to observe whether the actual run time matches the estimated run time is not a good idea. This can shorten the life of the lead-acid batteries. Also, unplugging the battery backup unit from the wall, which is how one might test the unit while utility power is available, can also harm unit due to the ground wire being disconnected, which in fact does stay connected during a power outage.

A Test button will switch to battery power, and measure that the output matches what was expected.

Lifecycle management

Batteries should be replaced in a single unit all at once. Conventional wisdom says installation dates (assuming new) should not be more than six months apart. Otherwise, the charging currents may be applied to newer batteries as the older ones continue to require it (since they recharge more slowly), which can shorten the life of the new batteries.

With most consumer products, we put batteries in and use the product until the batteries die, then replace the batteries. With a UPS on a business network, this is not how it should be done. When the battery fails in a UPS, it may or may not shut down spontaneously, which is the opposite of the intent. Or, it may seem to continue running, but when the power goes out even for a split second, it fails to provide power to the protected equipment.

In addition, UPS units handle more A/C power and have large batteries inside that are prone to sulfation and corrosion. They are thus more prone to spontaneous failure as they age than a desktop computer, rack server, or network equipment. Thus, battery backup units should be deployed on a time-based lifecycle. Battery backups should be replaced at most every three years. While some units may be warranted for longer, or you may have the option to replace only the battery while keeping the main unit, the entire unit should be replaced.