Ensuring continuity for your data center customers demands holistic risk management defined by rigorous risk protocols and analytical insight.
In this article, we set out five actions to achieve just this by managing data center downtime due to climate and natural catastrophe risks, specifically, we recommend you:
01
Data centers require different resilience strategies because, unlike telecom networks, which are typically decentralized and built with inherent redundancy, data centers are centralized, high-density environments. This makes them more vulnerable to localized disruptions such as power outages, natural catastrophes, geopolitical risks and cooling system failures.
Minimizing downtime starts with a nuanced and comprehensive view of the potential risks your data center faces. Ideally, you need a wide-angle risk and resilience assessment to evaluate both your exposure to all potential natural and climate risks, the likelihood of single and multiple-location impacts, plus proximity to critical utilities. You should also consider potential large-scale disasters that can cause damage across broad geographic areas.
Let’s say your data center is located in an area prone to hurricanes or earthquakes. To better manage potential downtime, you’ll want to identify and quantify the likelihood and magnitude of these events and their impact on your operations. Similarly, if your data center cluster relies on a single power source or water supply, you’ll need to understand the potential for outages and how they could impact operations and your ability to recover in the face of disaster.
A holistic data center risk and resilience assessment should also analyze not just the risks and impacts of natural catastrophes and utility outages but also other threats like proximity to high-risk locations or geopolitical risk.
Once you have this information, you can then rank individual or data center cluster vulnerabilities, which can inform estimations of potential downtime. You should make sure any vulnerability assessment is underpinned by your data center buildings’ characteristics, existing redundancies and mitigation measures and utilities dependencies.
Where you have multiple data center locations, assessing vulnerability scores and downtime estimates can help prioritize risk management efforts. By focusing on locations most vulnerable to downtime, you can allocate adaptation and risk mitigation resources more efficiently. And by having a combined risk score for multiple perils and multiple data centers within a cluster, you can better understand cumulative impacts and how you can manage the associated risks more effectively.
Where you have existing data center arrangements but are renewing with AI-focused servers, you can widen your assessment to focus on how key risks may increase due to increased heat production, power load requirements and weight.
02
Once you have a thorough risk and resilience assessment, you can enhance the physical infrastructure of your data center to withstand the most relevant natural catastrophes and climate risks. Depending on your location, threats and likelihood of impact, this may mean using earthquake-resistant building materials, protecting IT equipment with shock-absorbing mounts and seismic-rated racks, installing flood barriers and water detection devices, or building your data center to withstand high winds. Assessing these risks as early as possible means you can consider climate-resilience during the design stage and before construction
Use base isolation systems or dampers between buildings and foundations can absorb seismic energy, while using seismic-rated equipment can protect servers. Installing fire suppression systems and maintaining defensible space with fire-resistant landscaping around the facility can also help defend sites.
Depending on location, you can also consider installing flood barriers to prevent water ingress that could damage servers, power systems and cooling infrastructure; sealing all openings, such as for cables and pipes to protect equipment from the risk of humidity. You may also need to design facilities to withstand high winds.
You can supplement structural measures with advanced monitoring systems to detect and respond to threats as they happen. These systems can provide early warnings of potential issues, enabling you to take additional steps to minimize downtime. For example, if a severe storm is approaching, real-time monitoring and alerts could give you time to shut down non-critical systems and ensure your backup systems are ready to go.
03
There are many ways your data centers can achieve redundancy that ensures if one part of your network fails, another can take over without causing significant downtime. Consider combining different power sources, such as on-site generation, grid power, and renewable energy, to maintain uptime and reduce shared fate risks.
You could also build redundant infrastructure and data backups by setting up a secondary data center. To achieve redundancy and reliability, you can choose to build multiple data centers within a 100km radius near major internet exchange points or peering points. Such ‘clustering’ can provide redundancy by distributing workloads across multiple data centers. If one data center experiences an outage, the others can take over, ensuring business continuity. While clustering might enhance redundancy, it can also increase the shared fate risk of a single catastrophic event impacting multiple data centers at once.
Redundant cooling systems can help you maintain optimal temperatures in your data center, crucial for performance and longevity. Advanced analytics can help you not just with site selection, but to understand the intensity of heat and drought now and in the future and other environmental hazards. You can then link these hazards to your operational thresholds, allowing you to make informed and efficient decisions on cooling capacity, again helping to reduce potential downtime.
04
Implementing robust business continuity and disaster recovery plans is crucial for minimizing downtime and ensuring your data centers can recover from natural disaster or climate-related events quickly.
Choosing the appropriate design and loss return period (RP) is an important foundation of your business continuity and disaster recovery plans to minimize downtime.
A return period is a way to express how likely a climate event or loss is to occur in any given year. For example, a 100-year return period means there’s a 1% chance the event will happen in a given year. It helps quantify the frequency and severity of natural catastrophe and climate risks for data center operators, especially those serving financial services, healthcare, or AI workloads, are planning and shifting from 1 in 100 or 200 to a 1-in-500 year or higher return period.
This shift reflects the clustering of data centers in high-risk zones, the systemic impact of downtime across global operations and increasing regulatory pressure in addition to that from data center client expectations.
All this means data center design must account for climate change risks by shifting from historical data to future projections. It also means data centers require resilience measures such as redundant power and cooling systems, efficient water management, and increased building elevation or windspeed design to mitigate risks like heatwaves, flooding, wind and droughts. On top of the international and country-specific building design codes, standards such as the EU Taxonomy's Climate Risk Vulnerability Assessment (CRVA), the Code of Conduct for Data Centre Energy Efficiency and certifications such as the Building Research Establishment Environmental Assessment Method (BREEAM) provide frameworks to assess risks and improve resilience.
Data center and telecom asset recovery planning are both subsets of business continuity planning (BCP), which focus on minimizing downtime and restoring operations after an incident. While telecom asset recovery planning is more geographically dispersed and involves securing networks and communication links, data center recovery prioritizes physical infrastructure, critical IT systems and data storage within a centralized facility, often using off-site backups.
Your plans should address what happens before, during and after an event and cover all aspects of your operations, from the physical infrastructure to the data and applications you manage. In particular, your business continuity plan should outline steps you’ll need to take to keep your data center operational during a disaster, for example, establishing a secondary data center in a different location which isn't climate-exposed and that can take over if the primary data center is affected. Your plan should also define the procedures that will maintain communication with your clients and employees.
Your disaster recovery planning should include establishing a detailed inventory of all your equipment and systems, along with a plan for replacing or repairing any damaged components. It should also feature procedures for restoring data and applications, ensuring your clients can access their information as quickly as possible.
Crucially, you should have a plan in place for how each data center will respond to a disaster and how they will support each other if needed.
05
Enhanced inter-location communication can better protect your operations from natural catastrophe and climate events, in particular those impacting multiple locations within a cluster.
This is about establishing clear lines of communication between all your data centers and a system for sharing information in real-time. For example, you could use a centralized communication platform all your data centers can access, or set up a dedicated emergency response team to coordinate communication across multiple locations.
Regular drills and simulations can test your communication and coordination plans between data centers, ensuring they’re effective and all teams are well-prepared for any eventuality.
To help protect against natural catastrophes and climate risks, we work with risk managers, operational and sustainability teams, and engineering teams to ensure that loss modeling considers factors such as clustered exposures, climate-related perils, and downtime vulnerability. Contact our climate and natural catastrophe risk specialists to find out more.
