|
 |
|
General Information
|
Add To My Personal Library |
April 24, 2009
• Vol.31 Issue 13
Page(s) 28 in print issue
|
Monitoring The Data Center
Streamline & Prioritize The Monitoring Process In Order To Save Time & Money
|
With so many variables in a data center, it’s often tempting to monitor as much as possible, from hourly cooling levels to log reports to UPS consistency. But similar to trying to pay attention to a house’s doors, windows, appliances, and temperature, there’s only so much that can be handled without a feeling of being overwhelmed. Here are some top tips on what to monitor and when.
 Understand What’s Currently Being Monitored
Most data centers are very focused on monitoring applications and the IT equipment on which those applications run, and that makes sense, notes Michael Hogan, global offering manager for site and facilities services at IBM.
But the IT equipment only uses about 40% of the power that the data center consumes, he points out, and monitoring systems that use more than half the data center’s power are often overlooked or not integrated into a comprehensive view of the total data center operation.
If there is monitoring being done on infrastructure such as chillers, lighting, and power, it often happens that the two areas aren’t linked together to see how they might be interconnected. For example, a server may be monitored for its usage levels but not its power draw.
Another factor to consider when integrating a piece of equipment into a monitoring strategy is whether there are any existing intelligent controllers that can be utilized, which are sometimes built into newer equipment such as cooling racks.
Get A Holistic View
With a move to “green IT” equipment, overtaxed power grids, and an industry shift toward reducing carbon emissions, it’s likely that more data center managers will be asked to take a more holistic view of energy use, says Glen Siemetz, senior director of data center strategy and portfolio at Siemens IT Solutions & Services, a business unit of engineering firm Siemens AG (www.siemens.com).
Holistic thinking will also help at the data center level, adds Hogan, who notes, “In order to get a holistic view of the data center, it’s suggested that both the IT and facilities are monitored and combined into a single ‘dashboard’ with metrics that are relevant for the specific site.”
There should be a common power monitoring platform for IT equipment and building infrastructure components, Siemetz suggests, in order to baseline usage and improve data center efficiency.
Use Industry Standards
“Energy usage measurements, like the Green Grid’s PUE (power usage effectiveness), is a great step forward in this direction because it requires a monitoring view of building power usage, as well as IT equipment power usage, to determine the data center’s energy effectiveness,” Siemetz says.
“The PUE or DCIE will provide a means to compare efficiency over time,” says Hogan. “It will drive the basic requirement to understand where power is used and by what systems.” He adds, “The level of detail for monitoring the various systems in a data center will depend on the availability of data on specific pieces of equipment and the budget for implementing the monitoring system.” IT managers should take into consideration the recommendations of the industry groups such as The Green Grid, he believes.
Watch Consumption At The Server Level
When tracking power usage, many data centers look at overall consumption or examine usage at the server level, but it’s helpful to get even more granular than that, notes Motti Tal, executive vice president of marketing and business development at OpTier (www.optier.com).
Akin to cutting out unnecessary expenses by going through a budget’s line items, server power consumption can be reduced by examining compute cycles or even usage by various departments. “It’s all about optimization, particularly in this economic climate,” says Tal. “If you don’t understand usage—both by employees and in terms of power consumption—it’s more challenging to become efficient. Servers are often over consuming, and a data center manager might not realize it.” 
by Elizabeth Millard
Bonus Tips
Work with facilities management. Because automated dashboards can tie in facilities monitoring with IT equipment, it’s advisable to create a stronger relationship with a building’s facilities management department, notes Michael Hogan, global offering manager for site and facilities services at IBM. The goal is to link everything into a single view, including UPSes, servers, CRACs (computer room air conditioners), and other equipment, and it’s likely that a facilities manager is aiming for efficiency just as much as a data center manager. Evaluate what type of monitoring is done by that department and whether the two systems can be bundled together.
Don’t forget security. One area that will likely not be mixed in with everything else is security and, specifically, monitoring of firewalls and other intrusion prevention and detection systems. Although any appliances or applications can be blended within a general monitoring system, Glen Siemetz, senior director of data center strategy and portfolio at Siemens IT Solutions & Services, a business unit of engineering firm Siemens AG (www.siemens.com), believes that security warrants its own daily monitoring, at multiple levels. He says, “The security area of IT consists of a continual stream of attack and counter-attack prevention escalations that are enabled through monitoring. It reminds me of the Warner Bros coyote and road runner cartoons.” |
Most Useful Tip: Know The Daily Monitoring Points
According to Glen Siemetz, senior director of data center strategy and portfolio at Siemens IT Solutions & Services, a business unit of engineering firm Siemens AG (www.siemens.com), data centers have an array of points to check on a daily basis, including power, environmental controls, physical security such as alarms and locks, heating, ventilation, air conditioning, and fire detection and suppression.
For IT equipment, daily monitoring should be arranged for networks, security, servers, mainframes, storage, backup, and application components. He says, “Both building and IT monitoring are used to feed systems to measure faults, usage, capacity, performance, availability, and security. |
Time-Saving Tip: Get Automated
Although it may seem as if there’s far too many points for a data manager and IT staff members to track every day, it’s possible with automated controls that bring all the components together, Siemetz notes. There have been many developments and innovations lately in creating dashboards that can bring together both facilities and IT assets and create reports that give a more holistic view of a center.
In the past, monitoring response systems delivered only simple alerts and warnings, often issued after something went awry, such as the breakdown of an air-conditioning unit or the disruption of a network. But intelligent controllers are in many pieces of equipment and can be tied in with software for better insight and monitoring.
One example of this is data center cooling, Siemetz notes, where there are now cabinet sensors that are tied to a cooling system with variable speed fans. The fans adjust automatically to load and cooling requirements, based on cabinet sensor information. He says, “This cooling adjustment can save thousands of dollars in wasted data center energy on a monthly basis.” |
|
|
