Lessen The Effects On Your Budget By Keeping Older Equipment Humming

Key Points

• Inadequate power and cooling management can turn aging equipment into doorstoppers. Focus on cleaning up the environment first to squeeze the most life out of your hardware.
• Software patches and updates are just as important regardless of age. Don’t ease off on the maintenance regimen just because a server is past its best-before date.
• Equipment failure is often mistakenly attributed to age, when in many cases, sketchy network performance is what starts it all. Monitor surrounding infrastructure to pinpoint root causes.

Budget-strapped companies facing falling revenues and profits, and questionable futures continue to hold the line on IT budgets. Nowhere is this playing out more starkly than in the data center, where equipment is being pushed beyond its originally intended lifespan. Cost savings associated with avoiding capital investments are often more than offset by the challenges of keeping aging equipment working—and reliable.

“Your data center equipment is just like a car,” says Jeremy Littlejohn, chief analyst at RISC Networks (www.riscnetworks.com). “The more miles you have, the more important the maintenance.”

Unfortunately, as the miles pile up, increased maintenance needs often collide with growing business needs. Aging equipment could expose the business to reduced availability due to more frequent failures and greater difficulty in sourcing spare parts. Consider the following problems as you try to make do with what you have.

Too Much Heat

Heat is hardware’s worst enemy, and over the years, elevated temperatures—especially in small spaces—can reduce the lifespan of data center equipment. Popular methods for minimizing cold air loss throughout the facility are using blanking panels in server racks and installing air-locking grommets in raised-floor panels.

“An effective HVAC system is a must to extend the life of data center equipment,” says Peter Wallace, CIO for the City of Chesapeake, Va. “Organizations should make sure their HVAC system is supported with backup power so that temperature and power spikes do not impact equipment.”

Invisible Uninterruptible Power Supplies

UPSes are often mistakenly overlooked as a root cause of data center downtime. Hardware that’s plugged into an inadequately maintained UPS—or, worse, no UPS at all—will fail more often than equipment that is fed clean, monitored current. Older equipment is even more susceptible to fluctuation-related failure.

“UPS maintenance is very critical,” says Keith Gislason, director of product management for Avocent Products and Services, a division of Emerson Network Power (www.avocent.com). “You don’t build a data center unless you want your servers to stay up, and to do that, you need clean power. Maintaining your UPS prolongs its life and the life of your equipment in addition to reducing downtime. This is not an area that any business will ever compromise.”

A study on UPS reliability conducted by Emerson Network Power showed the MTBF of a UPS that received two preventive maintenance visits per year is 23 times better than a UPS that is not maintained.

Unmonitored Power

If badly maintained UPSes are like ticking time bombs, unmonitored power can similarly make attached hardware old before its time. Conventional power distribution units don’t provide adequate visibility into power usage. This can lead to power imbalances within the data center, which, beyond inefficient energy use, can also upset the cooling environment. To combat imbalances, IP-capable PDUs can be retrofitted to existing server cabinets and existing PDUs to more effectively balance power demands with actual supply.

“You will continue to provide reliable power distribution with the existing PDUs,” says Julie Brown, marketing manager with Server Technology (www.servertech.com), “but you will combine it with local and remote power and environmental monitoring via IP using the additional device.”

Too Much Hardware

As equipment ages and reliability drops, maintenance needs increase tremendous-ly. Instead of looking for best practices to streamline fixes to the same old hardware, Wallace recommends seizing the opportunity to get rid of the hardware altogether. Virtualization reduces the size of the hardware pool and its associated maintenance load.

Wallace says virtualizing helped the Chesapeake city administration avoid new capital expenditure by bypassing new hardware purchases and a data center expansion. The initiative extended their existing equipment’s lifespan and returned an annual $200,000 savings in hardware costs.

Slipshod Application Maintenance

Maintenance becomes more crucial to maximizing uptime as equipment ages, and the same is true of applications. IT must work with its business partners to define both acceptable and unacceptable levels of performance. This ensures proper prioritization of effort in the event of an outage and ensures that routine maintenance activities aren’t forgotten, either. Wallace recommends developing benchmarks for business performance that include applications as well as hardware. Littlejohn is more succinct.

“First, forget about hardware; software is king in the data center,” Littlejohn says, adding that most key forms of automation are more dependent on software than on hardware. “So even if you don’t update your hardware, make sure the software, BIOS, patches, etc., on your infrastructure and servers are all up-to-date.”

He also recommends keeping a close watch on application and event logs.

WAN Performance Ignored

When resolving data center performance issues, IT professionals often focus too closely on the facility instead of the network that feeds it. Littlejohn describes WANs as “the forgotten component” of data centers.

“I would say that in easily 25% of the data centers we evaluate, the WAN is not provisioned properly and shows performance issues,” Littlejohn says. “These are quickly blamed on the ‘old’ servers or ‘ancient’ switches.”

By focusing root cause analysis on the underlying network infrastructure, problems previously classified as hardware-related stand a greater chance of being properly categorized and resolved.

Article by Carmi Levy for .