Global Warming and IT

By: George Spafford

August 3, 2006

Global warming is a very threatening reality. Predicting how long it will last or how to gain the upper hand to stop its devastating progression we must first recognize that we’re part of an ever evolving world. Technological success or failure is directly impacted by rising temperatures. Consider the number of physical and economic reasons or effects it has on firms and why we need to explore some of them.

Excessive heat can wreak havoc on equipment and personnel. Most server rooms and data centers have air conditioning (AC) units installed to reduce temperatures to acceptable levels based on the requirements of the equipment used. Some data centers are running so much equipment that their cooling needs are substantial and a failure in the AC system can be catastrophic in terms of either outright failure or higher than normal component failure rates.

As a result, we’ve learned the need to have dedicated AC systems and monitoring to trend temperature and humidity levels and generate alarms when thresholds are exceeded. Groups in high reliability situations have even built in redundancy on their AC units and have power backup to ensure that the data center doesn’t bake during a power outage. Ensuring cooling during a power failure is often overlooked and can result in an unplanned shutdown due to overheating.

The problem we face today is that will see more risks manifest themselves given the warmer environment. As temperatures rise, AC systems will have to work harder than ever to lower temperatures to acceptable levels. This means that power requirements are going to increase. Moreover, large AC units have to dump their heat into the atmosphere and as the surrounding air that is already hot is heated further, efficiency will decrease. If you’ve ever stood next to cooling systems running you will notice they generate a tremendous amount of hot air. For groups running self-contained portable/temporary AC units, they should ensure that the exhaust is vented to the outside of the building otherwise they will be heating the very air the system is trying to cool.

As the demands for power increases, then the march to higher prices will be inevitable. This will affect budgets as actual expenditures may well exceed the budgeted amounts. As a result, if relationships don’t already exist with the local power company, contacts should be developed and pricing forecasts discussed plus an understanding of what they are doing to address power requirements in the area, their continuity plans, how calls should be escalated and so on.

With more air conditioning running causing an increased strain on the power network, there is also an increased likelihood of power grid failure. A transformer may blow, a line may burn out, etc. Moreover, the power company may be forced to implement rolling blackouts or even emergency shutdowns to protect equipment. Bear in mind that sizable portions of the power infrastructure were not designed for the consumption levels we are encountering today. For a whole variety of reasons, and cooling being a major one in the summer, we are using a tremendous amount of power.

The hotter temperatures will directly impact the power utilities. As they try to react to the higher demand, there will be accusations of poor planning when scenarios related to global warming have not been sufficiently thought out. While it is easy to pin blame, to be quite fair it is hard to plan for something that you’ve not encountered before. Once it happens, you know what to expect and how to prevent or recover a particular type of incident from happening again.

As an interesting note, the AEP Cook Nuclear power plant had to take a unit offline during the hottest days of the summer because the water from Lake Michigan was too warm to cool the reactor to levels within design specifications. As a result, Cook must investigate using far more expensive ground water to cool the facility. While it is cheaper than having the unit sit idle, it is also something that will be costly over time. If they, and other utilities forced to do infrastructure upgrades, are unable to pass along the costs, they will be in the difficult position of not being able to afford further improvements or, in the worst case scenario, they would not be able to stay in business.

IT needs to plan for ways to reduce temperatures in data centers. The following are some of the actions that are possible:

• Perhaps the easiest, make sure that unnecessary equipment is not running. This includes monitors. Investigate remote power switches if need be.
• Review what power management capabilities the systems have relative to business needs and then develop policies regarding power management settings.
• Investigate using cooler running power supplies. It is reported that Google pays a premium for power supplies that run cooler knowing that they will recover the costs by lower electric bills.
• Understand what your ideal temperature band is. It is not cost effective to cool a room more than what is needed.
• Have a professional who is experienced in data center design and cooling review the current layout and see what options exist to improve efficiency and effectiveness.
• If a building roof is black, investigate having it painted or sealed using a white finish to reflect heat vs. being a giant solar heater.
• If an AC unit hasn’t been serviced in a long time, have it inspected to ensure it is running as efficiently as possible. There may be filters that need to be cleaned or replaced, belts tightened, refrigerant levels corrected, and so on.
• Verify the capacity of the current cooling systems and ensure that it will meet current and planned requirements. Some of the systems may already be running at, or beyond, capacity.
• Ensure all data centers, server rooms, and other critical areas have temperature probes installed. Be sure to purchase and install them with both alerts and trending to avoid heat related incidents caused by either cooling system inadequacy or failure.
• Review the temperature trend information and take it into account when performing capacity planning and budgeting for the following year.
• Review continuity plans and verify that there is sufficient cooling capacity during outages. This includes ensuring that the systems will have adequate power as well and not shut off when power is lost.
• It is easy to brainstorm ideas and then rank them according to risk and cost effectiveness.

For final consideration, with the higher air temperatures melting glaciers and warming waters, it is predicted that the number and severity of storms will increase. Gulf and coastal states are watching this closely. The resulting weather will affect not just environmental controls in the data center, but the continuity of operations for organizations and entire economies. Having current and tested business continuity plans in place is always a wise idea.

There aren’t any easy answers on this topic. Groups need to plan and mitigate their risks. Not only do we need to react and deal with the threats that it introduces but we must also think how, as nations, corporate citizens and individuals, we can reduce global warming.

 -----

Note: To stay abreast of global warming, consider subscribing to the author’s newsletter. More information is available at http://www.spaffordconsulting.com/dailynews.html

Technically, the units are referred to heating, ventilation and air conditioning systems, or “HVACs”. As the article is focused on global warming and cooling, the emphasis has been placed on the “air conditioning”.

It is advisable to monitor temperature and humidity. Many of the network enabled probes feature both temperature and humidity sensors that then report the data back via SNMP. In this article, the focus is on temperature but during reviews of climate control, be sure to understand required temperature and humidity levels both of the equipment and the media in the data center.