Knowledge Base / News
January 23, 2015
(View Original Article)
Processor – Vol. 37 Number 2
The many benefits of monitoring environment conditions in data centers have been fairly well documented in recent years. Still, many enterprises fail to implement even basic monitoring until after actually experiencing a problem.
Though some reasons why enterprises take a “later than sooner” approach are easier to justify than others, realistically every company should be performing basic environmental monitoring, if for no other reason than the peace of mind that comes from “knmowing all is well when a proper system is in place and there have been no recent alerts,” says Michael Sigourney, president of AVTECH (888.220.6700; www.AVTECH.com). Here are some strategies for where to get started or how to boost your current monitoring.
With environmental monitoring (EM) products readily available and so much evidence detailing their benefits, why do some enterprises fail to implement them?
Time and money are typical reasons. “Although EM products are inexpensive, given the exponential return on investment, it’s often easier to think products that users touch and work with regularly are more important,” Sigourney says. Some enterprise leaders believe that once an environmental disaster occurs, it won’t happen again. Without preventative measures and ongoing monitoring, however, they likely will, Sigourney says.
Ken Koty, sales engineer at PDU Cables (866.631.4238; www.PDUCables.com), says that some data center managers also believe that once commissioned, newer data centers with lower densities and new equipment don’t represent a significant risk for problems. This changes over time, however, and eventually events will occur, he says. Other enterprises simply resist change and the learning necessary to implement EM systems, especially if things are currently going well.
Aheli Purkayastha, product manager at Purkay Labs (774.261.4444; www.PurkayLabs.com), says when you consider how many servers, racks, aisles, and sensors an enterprise potentially has, it can seem daunting to install, configure, and manage EM systems. Installing an EM system may seem tedious or unnecessary if problems seem minimal or small enough to ignore. For smaller enterprises, installing and sifting through EM-related data can be challenging, she says. “Barring a crisis, the tendency is to continue to avoid monitoring the environment,” she says.Everyone Can Benefit
The wait-and-see mindset of some enterprise leaders is changing as the benefits of EM receive more attention. Sigourney says attitudes will further shift as monitoring via cloud services increases. More enterprises are measuring PUE and recognizing their reliance on data centers for mission-critical data, which is also helping drive awareness. Sigourney says any facility can benefit from EM, especially with solutions starting as low as $145 and the extreme impact an event could pose.
Purkayastha says although data centers vary in size and cost, all “house environmetally sensitive equipment require the same basic measurements to ensure that servers will function.” For example, EM helps ensure that inlet air entering a server meets equipment or ASHRAE standards and helps prevent failure or downtime, she says.
At the same time, Purkayastha says, inlet temps may fall within an acceptable range, but the enterprise is still wasting money and electricity on unnecessary coolin. For example, rather than detect and fix a hot spot through monitoring, IT might just turn down CRAC settings to pump an even colder air supply to the server inlet. “The root cause of the hot spot will continue, even though the inlet temperature might be OK,” Purkayastha says.
Koty says monitoring for high humidity levels can help prevent corrosion of electrical components and that monitoring for low humidity levels can help prevent issues with static electricity. Liquid detectors are often overlooked, he says; he recommends placing such detectors inside cooling equipment, near potential leaks from nearby pipes, in areas where water would result from a flood/disaster, and in air-conditioning condensation trays to detect overflow.
Cover The Basics
At a minimum, you should monitor for humidity, dew point, power loss, power quality, differential pressure, and inlet temperatures. In addition to inlet air temperatures, Purkayastha recommends taking measurements across the top, middle, and bottom of server racks to note temperature variance and across the aisle to account for variations. Also measure how loads vary throughout the day over an extended period.
Most enterprises begin with monitoring by measuring temperatures or addressing a condition that recently cause a pain, Sigourney says. He recommends starting EM with key concerns in the facility and expanding as budget and time allow. Budget each quarter to add more monitors and sensors for new or ongoing concerns, ensuring the chosen platform enables expansion. Avoid software or services that charge more as additional monitors are added, and seek flexibility to monitor from anywhere and anytime from any computer or mobile device, he says.
Koty encourages monitoring UPS and battery systems, backup diesel generators and their fuel supply levels, room entry points, and all critical switch gear and circuit breakers within them.
Don’t Forget Security
Tanja Lewitt, CEO of Kentix Innovatice Security (844.536.8492; www.Kentix.com), says while most enterprises are monitoring for physical access into data centers and server rooms, they aren’t necessarily monitoring access down to the equipment level. “We see more and more incidents at the equipment level,” she says, from disgruntled employees or someone looking to steal equipment. In general, monitoring for physical security access can involve motion, visual, and audiovisual aspects, Lewitt says. “We don’t like to talk about these things, but they’re real-world,” she says.
“Think about it: If you have even $30,000 or $50,000 worth of servers go down, the process this takes down with it could be worth millions.”Michael Sigourney, president of AVTECH (888.220.6700; www.AVTECH.com), says there are six reasons organizations should monitor their IT and facility environments: being able to know immediately when problems occur; reduce/eliminate downtime; maintain maximum energy and resource efficiency; protect/extend equipment, product, and property life spans; maximize reliability and peace of mind; and stay online and open for business.
# # #
It’s important to recognize the dynamic nature of data center environments, including the impact of seasonal changes. “Changes in weather and climate are so unknown now,” says Tanja Lewitt, CEO of Kentix Innovative Security (844.536.8492; www.Kentix.com). “Climate changes are subtly changing what’s happening with this equipment, and if you’re not monitoring the environment, you’re not going to know that,” she says.
Ideally, enterprises should be monitoring for server and CRAC delta temperatures to help understand the effectiveness of cooling, says Aheli Purkayastha, product manager at Purkay Labs (774.261.4444; www.PurkayLabs.com). “As you increase load capacity, add blade servers, or make any changes to the data center layout, this measurement will benchmark your cooling efficiency and act as a preventative measurement to any server failure/downtime,” she says.
Watch The Time
While managing data centers, Ken Koty, sales engineer at PDU Cables (866.631.4238; www.PDUCables.com), says he found that time-stamping the monitoring of temperatures, power to cabinets, and humidity levels can helpt detect server problems. Typically, when a problem does occure, he says, an IT person will call a vendor that will look first to place blame for the problem on input power, temperature, or humidity. Having time-stamped information available can help eliminate finger-pointing and get problems fixed faster, he says.