We’re entering the heat of summer, and that means we’re all thinking about one thing: cooling down. For data centers in particular, this is a huge concern. Overheated data centers can cause big problems for organizations large and small.
While servers typically shut down when they hit their 99 percent heat limit to prevent serious damage, these shutdowns can corrupt vital information or wreak havoc on revenue. For the typical organization, unplanned downtime costs $7,900 a minute on average. For a larger company like Amazon, it can be as high as $66,240 per minute. Additionally, years of overheated data centers can cut equipment longevity, forcing an organization to replace servers every two years instead of every five. These costs add up. As data continues to grow, both in volume and importance, it’s more vital than ever for organizations to take data center cooling seriously.
Most organizations have some cooling systems in place to keep their servers running, yet few systems are perfect. Many companies settle for good enough, but this can put information and revenue at risk. Instead, organizations should look into ways to ensure data center health that range from the quick and simple to the inexpensive and impactful to the robust and long-term.
Quick cooling fixes
If you work for a small or medium-sized organization, it’s likely your data center is also on the small side. For that reason, these simple quick fixes can make big strides in data center cooling.
- Blanking panels — These panels cover the front of unused rack space to improve airflow throughout a data center. They’re relatively simple to install, and can significantly help with temperature control in smaller centers.
- Spot cooling — Spot cooling is one of the most effective ways to cool down smaller data centers. Organizations can install small, portable cooling devices to handle hotter areas within the center.
- Reorganize hardware density — When equipment’s stacked or huddled together, it can significantly raise heat levels, but by simply distributing servers evenly throughout a data center, organizations can cut down on potential overheating.
- Retire idle IT equipment — The easiest way to reduce data center temperatures is by eliminating or consolidating unnecessary devices. By just getting rid of needless servers, organizations can eliminate excess heat and keep temperatures low.
- Cable management — Cable management is one of the most overlooked fixes to overheated data centers. Since servers expel heat from the back, sloppy cable work can trap excess heat and feed it back into the front of devices, causing them to overheat. Instead, streamline cable work to make sure hot air flows evenly through the center.
Inexpensive cooling fixes
There are also a number of ways to cool a data center that take a little more effort than the quick fixes above, but are relatively inexpensive solutions. Keep in mind that these fixes will have the biggest impact for small or medium-sized data centers.
- Service and maintenance on existing cooling systems — Just like people, data centers need checkups. By having your HVAC guy stop in a few times a year, you can ensure that your air conditioning is running at maximum efficiency and solve any potential glitches in air filtration, interconnections to the chiller on the roof, and so on before they result in a data meltdown. These checkups are relativity inexpensive and cost much less than system downtime or installing a completely new HVAC system.
- Virtualize hardware — Another inexpensive solution is virtualizing servers to cut down on the number of servers in use, and in turn, reduce overall data center heat. The amount of heat generated by virtualized hardware is less than the equivalent non-virtualized setup. The one caveat is that a virtualized machine can generate more heat than the server it’s replacing, increasing the heat density per virtualized rack and creating hot spots that have to be managed and exhausted. However, for data centers that have already reached their cooling capacity, virtualized servers reduce the overall heat footprint and allow the system to continue to operate without having further upgrades or expansions.
- Hot and cold aisles — Instead of positioning all of the rows in your data center to face the same way, reposition the racks so that they face each other. This prevents the hot and cold air from mixing and allows centers to better contain and expel heat. Rearranging your data center takes planning and investment, and involves some downtime while hardware is being moved, but in the end, you’ll improve your data center’s air flow and cooling.
- Exhaust heat — Knowing where your data center’s hot spots are and how to cool them is one thing, but effectively exhausting the heat from the room is equally critical to keeping a data center cool. Typically, hot air is pushed out through the ceiling, wall, or floor to a separate room — a plenum — that doesn’t contain any critical IT infrastructure. Exhausting heat from the data center will allow your cooling systems to work smarter, not harder, in order to maintain your data center’s ideal temperature.
Long-term cooling fixes
Long-term data center cooling fixes are the cream of the crop in terms of preventing centers from overheating, but not all companies need to undergo these large-scale overhauls. Usually large, Fortune 100 companies choose to undergo these long-term fixes since they have the most to lose in the case of downtime.
- Computational Fluid Dynamics (CFD) — These data center modeling simulations use fluid dynamics to discern heat patterns within a data center. These simulations can determine hot spots and predict where they could occur if heat continues to rise. After undergoing this simulation, companies can install cooling solutions, such as in-row cooling equipment, to reduce heat in high-risk areas.
- HVAC cooling projects — If all else fails, organizations can always undergo large-scale HVAC renovation projects to improve the power and efficacy of their cooling systems. These are often costly, but ensure the greatest cooling overall.
Too many organizations are not on top of their data center cooling. They often view it as an afterthought, and this can cause serious problems for data centers. Organizations need to take a step back and re-evaluate the cooling capabilities of their centers. By spending a little extra time tending to server needs upfront, they can save big in the long run.
For more information or additional recommendations to cool your data center, email Power@SHI.com.
Jason Yu contributed to this post.