In order to minimize data center outages and their associated impacts upon an organization, the environment for critical business applications must support the objectives of scalability, security, disaster avoidance and recovery, and—most important—high availability. The operating condition of your data center facility is part of what enables high availability. And the only way to ensure operating condition—and thus high availability—is through good maintenance practices.
When you think about the cost of the building’s infrastructure components and systems, not to mention the cost of the IT equipment, and the business value of the data supported, the cost of proper building-systems-infrastructure maintenance is small. And yet too many data center owners and operators fall shmaintenance practices, opting for the least costly, lowest-service-level contract, or attempting to perform certain maintenance tasks in-house. With the possible exception of filter changes, most other building-systems maintenance duties are best left to the professionals. Not only is the equipment expensive to repair should something go wrong during your maintenance attempt, but more important, the equipment can be dangerous and even pose a life/safety risk—either during maintenanceattempts, or over time—if maintenance is not properly performed.
Professional maintenance programs encompass a variety of offerings and are available not only from manufacturers’ representatives, but also from certified third-party vendors, some of whom are certified on multmaintenance vendors offer services that range from immediate response (on-site within 2–4 hours, 365 days a year) to best efforts.
In many cases, it is financially advantageous to negotiate a maintenance contract along with the original equipment purchase. Don’t just accept a one-year new equipment warranty as is. The warranty can often be negotiated to include routine maintenance and even to extend beyond the first year. As with a car, maintenance costs are minimal when the equipment is new, and repair costs should be zero during the initial years of operation. Thus, a manufacturer or vendor confident in the quality of their product should view extending the warranty and/or including initial routine maintenance as a very small concession toward securing your order—especially a large order.
While building-systems-infrastructure maintenance procedures can vary baseage, and other considerations, there are several prevailing best practices. These include monthly generator and HVAC maintenance, and quarterly maintenance of most other equipment. (See next page for a detailed maintenance schedule.) Small things can also make a big difference. Filters, fluid checks, and vacuuming are three of the easiest preventative maintenance tasks to tackle. Having a preventative maintenance contract in place will ensure that these and other maintenance tasks are accomplished in a timely manner. Failure to conduct proper preventative maintenance will significantly reduce the efficiency and life expectancy of your building-systems-infrastructure components. Case in point: batteries are commonplace items; they run our children’s toys, power our cell phones, and start our cars. But we tend to take batteries for granted … until they fail. The same holds true in our data centers. A few bad battery cells and the UPS can’t handle the IT load; a weak generator battery and the generator fails to start. Both scenarios produce the same outcome: a crashed data center with little or no ability to re-start and restore until the primary power is reestablished.
On the upper end of the preventative maintenance scale are infrared thermography scans. These heatdetecting (temperature-gradient) photographs can quickly detect weak circuits, relays, fuses, bus bars, breakers, cable connections, transformer coils, etc. By proactively replacing these weak, worn, or defective components during a scheduled maintenance window, you may avoid an unplanned data center outage, along with the associated headaches and costs.
Some helpful hints … if your data center’s air conditioners seem to require filter changes too frequently, consider vacuuming your under-floor air-plenum. This is an annual best practice. Dirt, dust, and debris, which collect in and are blown through raised-floor cavities, will eventually make their way into the A/C filters. Likewise, change fissured ceiling tiles for clean-room type tiles. Fissured tiles continuously flake, giving off microscopic (and larger) particulates, which also end up in the A/C filters—not to mention on circuit boards, on disk surfaces, and in IT equipment filters.
Finally, spend time with your maintenance contractor when they’re onsite. There are a thousand excuses for not monitoring your maintenance vendor—everything from “I’m too busy” to “I trust them.” It’s usually not a question of honesty, or time management, but one of relationship building. This maintenance contract will likely be in effect for years and you’ll probably see the same maintenance technician several times each year. If the technician sees you and believes you’re interested in the well-being of your data center and its building systems infrastructure, that technician is more likely to spend an extra few minutes and make an additional effort on your behalf. Go ahead, ask a couple of questions too. You might learn something about the equipment that supports your IT environment—and that added knowledge might one day serve you well.
The old saying, “Every 3,000 miles, or three months, whichever comes first,” has extended the operating life and increased the value and efficiency of many an automobile. Isn’t your data center at least as important as your car?
