How to Speak Like a Data Center Geek

Design for Data Center Resiliency to Reduce Downtime Risk

Plan for worst-case scenarios when designing new facilities or upgrading existing buildings to limit the impact on operations

Stephen Donohoe
Greg Metcalf
Andrew Higgins
Design for Data Center Resiliency to Reduce Downtime Risk

Just like any other building, data centers can be impacted by extreme climate conditions, such as excessive heat and flooding. What’s unique to data centers is the volume of businesses that require high availability of the infrastructure in those buildings. At Equinix, we have more than 10,000 customers worldwide, and all of them rely on our data center operations to help them serve their users without missing a beat.

Over the past several years, we’ve seen the variety and frequency of these worst-case scenarios increasing due to climate change that has resulted in extreme weather conditions such as heat, rain and wind. In 2017, wildfires came within three blocks of a northern California county’s data center. Data centers also experienced cooling system failures during record heat in the U.K. in 2022.

The design phase is crucial for data center construction. Our design teams must account for everyday and worst-case scenarios that pose the risk of causing downtime and then proactively build in the necessary resiliency. Taking these measures helps prevent damage to buildings, hardware and equipment and helps protect data.

In this blog, we’ll focus on the importance of using the design phase to plan how we manage two of the main threats to our Equinix IBX® data center operations: excessive heat events and flooding. We’ll also look at how the ever-changing dynamics of climate change demand constant monitoring and assessment to determine how to adapt our design strategies.

How digital leaders achieve sustainable digital transformation in uncertain times

Read how IT leaders deliver digital transformation and sustainability objectives simultaneously against a macroeconomic backdrop that is gloomy at best.

Download Now
How digital leaders achieve sustainable digital transformation in uncertain times

Design for local climate conditions and incorporate redundancy

 Extreme weather conditions, particularly those found in hot and dry environments, stress the ability of cooling systems to run at designed capacity. Designing for seamless data center operations during excessive heat events and incorporating redundancy are essential. To determine the appropriate target for ambient temperature, our design teams analyze historical extreme weather data, such as the return value of extreme temperatures, for predictions on what the highest temperature could be in the foreseeable future.

Improving PUE (energy performance) of data centers by raising temperatures from historically low operating temperatures is a do-more-with-less sustainability imperative. And, as average external ambient temperatures continue to rise, so have industry standards for safe operating temperatures in data centers. That’s why we like to say that warm is the new cool.

The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) recently updated its guidance,[1] stating that A1 (enterprise level) class equipment can now safely operate at temperatures as high as 27 degrees Celsius/80.6 degrees Fahrenheit. Previously, ASHRAE guidance was 66 to 77 degrees Fahrenheit.

The revised ASHRAE guidance means data center operators may be able to turn up their thermostats a few degrees to reduce total energy consumption without impacting the safe operation of IT infrastructure. While this may sound like a minor change, those few degrees can lead to resilience in the face of higher temperatures and a reduction in GHG emissions when applied on a global scale.

Cooling is essential to the safe operation of IT infrastructure. With climate change driving temperatures up, operators must select the right cooling system to mitigate the increased ambient temperatures in their data centers. We design for projected temperature peaks using both ASHRAE data and our local knowledge of microclimates in the specific locations where we operate, along with allowances for rising temperatures due to wind direction and adjacent plant. We also ensure that chiller plants can continue to operate when outdoor conditions are even higher than designed. Doing so ensures continued cooling inside the building, albeit possibly at slightly higher than the intended internal design conditions, for the short duration of the above-extreme outdoor conditions.

Heat rejection equipment inevitably needs to be maintained, and ultimately periodically replaced as it ages. This presents the opportunity to select higher efficiency equipment, accommodating future technological innovation, and even higher temperatures than currently forecast if it proves necessary. In other words: design to be flexible over time.

In addition to our cooling system design approach, we incorporate redundancy to help ensure that the designed cooling capacity is always available in the event of planned equipment maintenance, unplanned failure or capacity degradation. Additional redundant heat rejection plant is also a safety net should local conditions exceed the design climatic conditions allowed.

Use maps and historical data to lower flood risk 

Increased flooding is another result of climate change. Managing this risk starts with the site selection of new data center locations. Where we plan to locate a data center determines our due diligence process. For example, in the U.S., we have access to current maps produced by the U.S. Army Corp of Engineers, considered the de facto standard for flood risk management. These maps don’t exist in other countries, requiring us to conduct hydrological studies. Engineers study historical data to understand the potential for worst-case flooding events in three categories:

  • Fluvial floods occur when rivers overflow their banks due to heavy rainfall or melting snow, such as what we’re seeing in California due to the massive snowpack that accumulated this year.
  • Pluvial floods, also known as flash floods, occur when localized heavy rainfall, like a Mumbai monsoon, overwhelms the drainage capacity of the land and causes rapid flooding.
  • Coastal floods, also known as storm surges, occur when strong winds and low atmospheric pressure push seawater onto the shore, as we’ve seen when hurricanes progress up the East Coast of the U.S.

The historical frequency of these three flood categories determines the potential return period for a flood. Using historical data to predict the potential for flooding is similar to studying the return value of extreme temperature data, as referenced earlier.

Once the research is complete and the design team has selected a specific site, we start identifying requirements that will help protect the building in the event of a flood. For example, maps or hydrological studies may recommend how high above the predicted flood level the building should sit. If the planned location for the building has the potential for pluvial floods, the building design must ensure proper drainage. Doing so will avoid rainwater pooling to surround the building with water after a heavy rainstorm. In the case of coastal flood risk, installing barriers to block incoming seawater is built into the design plan.

Apply data center design best practices

Data center design is integral to ensuring data centers function well on a daily basis, even in worst-case scenarios. In nearly 25 years of designing data centers for Equinix, we’ve established a few best practices for protecting our buildings and everything inside:

  • Incorporate climate risk assessment, such as extreme temperatures, flood risk, wind risk, drought risk and water stress, as an integral step in your due diligence process for site selection and identify mitigation measures during the design process.
  • Partner with your internal risk management team to gain insights on climate change trends, new regulations and other factors that help guide design decisions.
  • Design for future facility upgrades with the least possible disruption to data center operations.
  • Design air-cooled chiller systems for localized climatic conditions, such as wind direction and heat sources, and recirculation mitigation.
  • Design buildings for indirect cooling to avoid using outside air, which contains pollutants, dust, dirt, debris and moisture that can accelerate server equipment fouling or aging.
  • Consider your shell construction strategy carefully by location; means and methods of construction vary highly globally. A modular approach can make sense for smaller deployments, as discussed in a recent blog on modular data centers.

To learn how data center design trends align with sustainability, read our Sustainable Digital Transformation Guide.

 

[1] Data Center Temperature and Humidity Guidelines, TechTarget.com, April 29, 2022

Avatar photo
Stephen Donohoe VP, Global Data Center Design
Avatar photo
Greg Metcalf Senior Director, Design
Avatar photo
Andrew Higgins Director, Engineering Development & Masterplanning
Subscribe to the Equinix Blog