Executive Summary
- Data centre resilience has largely been designed around a centralised model where critical IT is concentrated in dedicated facilities.
- Edge computing challenges that proven model – requiring a fundamental rethink of resilience and the evolving role UPS systems play in protecting increasingly distributed critical infrastructure.
- Chris Cutler, Data Centre Business Development Manager for power protection specialist Riello UPS, explores the shift towards edge computing.
When we talk about edge computing, the conversation usually focuses on latency, data volumes, and the need to bring processing power closer to where the data is generated.
But perhaps there’s a fundamental question we’re missing: what happens to power protection when critical IT infrastructure isn’t located in a purpose-built data centre?
Data centre resilience has largely been designed around a centralised model where critical IT is concentrated in dedicated facilities featuring resilient electrical infrastructure, UPS systems, generators, and cooling, supported by specialist engineering teams.
Edge computing challenges that proven model. Processing is distributed across offices, retail sites, manufacturing plants or factories, telecoms hubs, transport infrastructure, and other places that were perhaps never originally designed to operate as a data centre.
Uninterruptible power supplies remain one of the key layers of defence against disturbances to the electrical supply. But when you distribute critical infrastructure across potentially hundreds or even thousands of edge locations, resilience cannot solely be thought of in terms of protecting an individual site. The way we think about power protection – and resilience in general – needs to change.
UPS Still Matters… Maybe More Than Ever
Whether an operator’s installation consumes several megawatts or a few kilowatts of power, one of the fundamental roles of a UPS doesn’t change – protecting sensitive electronic equipment from disturbances to the mains supply. But the size of the load tells us very little about the criticality of what it supports.
A UPS maintains continuity while the normal power supply is unavailable or outside of acceptable limits – that could mean providing enough battery runtime for the generators to start or giving IT systems time to shut down in a safe and controlled way.
In addition to protection against blackouts, online double-conversion UPS topology can also regulate the power to connected equipment and helps decouple the load from many disturbances on the incoming mains supply. This power conditioning capability can be particularly important at the edge, where equipment may be connected to potentially less controlled electrical environments than a traditional data centre.
The tried and tested approach to data centre resilience is to build increasingly resilient facilities and architectures – concepts such as redundancy immediately spring to mind.
But trying to replicate conventional data centre resilience architecture across multiple edge locations could create significant cost, space, and overall operational challenges.
A compact edge installation might not have space for redundant infrastructure. The appropriate battery autonomy may also be very different from that of a larger data centre, depending on the application and wider power architecture. And there’s a decent likelihood there won’t be a physical presence onsite able to intervene when something goes wrong.
However, that doesn’t mean operators should accept lower resilience, it simply means asking a different question: how much resilience does the workload require, and where is the most appropriate place to provide it?
For some applications, uninterrupted local operation will be non-negotiable, while in other settings, a controlled shutdown may be acceptable, or it might even be possible to move workloads to another location.
While the answer will depend on the specific application, the principle remains consistent – design resilience around the workload.
From Site-Specific To Estate-Wide Resilience
The individual site remains the first layer of resilience. That involves choosing a UPS that’s appropriate to the critical load and physical environment, with suitable battery autonomy. In edge settings you need to consider factors that are less influential when you’re talking about a purpose-built data centre, namely ambient temperature, space, ventilation, accessibility, and quality of incoming electrical supply.
But edge computing also introduces a second layer – the resilience of the entire estate. An organisation running 50 edge sites, for example, doesn’t just have 50 UPS systems. That’s 50 critical power assets that need to be monitored, maintained, and managed, potentially spread across hundreds of different physical environments.
Whereas at a conventional data centre, engineers are typically closer to the equipment and can inspect it more readily, at the edge it may not be practical or economical to send someone to investigate every alarm or fault.
Remote UPS monitoring therefore becomes an increasingly important part of the power protection strategy for edge installations. Monitoring battery health, voltage and load levels, temperatures, and alarms gives operators invaluable insight into their critical infrastructure before issues can develop into more serious outages.
Such information is useful for one site, but across an estate comprising hundreds of UPSs, it becomes a much more powerful operational tool.
That opens the door to a more proactive, data-driven approach to maintenance, complementing scheduled preventive maintenance with insight that helps operators identify where intervention is most urgently required.
Getting Resilience Right
We do need to avoid the mistake of going too far in the other direction though. Just because an edge site is physically small doesn’t mean operators can under-engineer their power protection. It goes without saying that a low power installation can still support a highly critical workload.
Resilience needs to be layered. So at the local level there’s the UPS providing protection against interruptions. At the estate level, remote monitoring and management make it possible to manage large numbers of locations as a coherent infrastructure rather than a collection of isolated installations. Then at the workload level, redundancy, failover, and prioritisation can provide another layer beyond the physical site.
This is possibly one of the biggest changes edge computing brings to critical power. Resilience doesn’t mean making every individual site a ‘fortress’. Instead, it’s about ensuring the overall service remains available where the application architecture permits it, even when individual sites, UPS, or power sources eventually fail.
Defining Resilience For The Edge
Data centre owners and operators have spent decades perfecting the resilient building. But edge challenges the industry to think beyond the facility itself.
The UPS remains central to the overall power protection equation, because distributed compute still mainly depends on a stable and reliable electrical supply. However, its role now sits in a wider setting where local power protection, remote monitoring, data-driven maintenance, and workload resilience all go hand in hand.
For edge operators the question shouldn’t be whether every site can survive every possible failure. It’s whether the overall infrastructure can continue delivering the required service when an individual site or UPS fails.
For distributed workloads, that’s an important definition of resilience. And as more critical computing moves beyond the walls of the traditional data centre, it may become the definition that matters most.
