If energy transition is the future – why are data centre power designs stuck in the past?
Energy transition is both a global and local phenomenon. As such it means different things to different people. For oil and gas companies it means moving away from fossil fuels. For those in the business of generating power, it means a shift to renewable energy resources (RERs).
What transition means to those developing new and running existing large data centres is thinking differently about future power system designs.
It extends to re-evaluating whether existing designs are fit for purpose over their original 20-25 year intended lifespan.
In simple data centre operational terms, transition means addressing how waste can be cut through more efficient use of the energy supplied. This requires new thinking about how power chains are being designed and operated.
For many, the idea of a fundamental change in power system design is a concern – these are fixed designs which are both complex and costly to change. But there is an approach which does not have to mean huge risks, disruption or expense.
Adaptable Redundant Power (ARP), a patented system also known as Power as a Service is an innovative approach which can free trapped or stranded power capacity, provide flexible power to where the IT demand is greatest and help maintain critical applications in the event of an outage.
ARP tackles the issue that in traditional design deployments, each data hall can only access the power infrastructure with which it is associated. With ARP, the power chain itself becomes flexible.
ARP makes data centre power supply adaptable for a range of applications
ARP has different operating modes which do different but complementary things. These are: Adaptable Redundancy, Inherent Redundancy, and IT Load Prioritisation.
ARP mode 1 is Adaptable Redundancy. This bridges the gap between flexible IT workload SLAs, and a static power design that is permanently fixed to one SLA.
It gives operators of data centres the ability to flex the power provision to the changing IT load requirement – using the same power system infrastructure.
Adaptable Redundancy addresses the issue that the power system has a single SLA supporting a variable and mixed SLA application environment. It does this by dynamically regulating power system availability to align with variable IT SLAs.
Inherent Redundancy (IR) mode 2 is the ability to access trapped power and divert it to provide virtual redundancy. ARP IR Mode produces significant CAPEX savings. IR provisions and manages predetermined redundant levels to the IT Load whereby redundancy is derived entirely from unused power capacity. IR can also enable unused power capacity to be accessed during normal operating conditions.
IR addresses a common industry challenge of power capacity becoming trapped or stranded and consequently wasted, during day-to-day IT operations when equipment is added, changed, or moved in data centre racks.
ARP’s IT load Prioritization mode ensures that during an outage any remaining power in the system is directed to keeping the most critical applications operating and available to avoid a data centre outage.
Data centres are and will continue to be designed as physical built environments which are subdivided into data halls.
A single power system topology assigned to only an individual data hall makes neither economic nor environmental sense. Should expensive equipment and power lie idle and wasted at a particular physical location waiting for a load to arrive when the service it provides – electricity – can be directed to different partsof the facility?
ARP works by capturing and then using unused power capacity in the data centre. By providing higher utilization of the power system, it delivers significantly improved returns on invested capital.
To date, within existing large facilities, the main barrier to changing how power is provisioned has been the risk associated with the disruption and vast expense of ripping and replacing large parts of the physical switchgear, UPS and other infrastructure.
In a choice between changing the power chain topology or keeping an existing set up (despite its inflexible power delivery, stranded capacity, low utilisation and outright waste) the prevailing view was ‘if it ain’t broke…’
This is because alternative choices were terrible.
However, ARP provides a way for existing data centre power systems as well as those being designed for new developments to be made both flexible and responsive, driving efficiencies and lowering capital and operating costs. This can be done using existing equipment – in fact driving higher utilisation from existing investments.
The infrastructure ecosystem of commercial, enterprise and cloud data centres is continually adapting to meet the changing requirements of IT. Power topologies must be adapted for the new realities of energy transition.
The direction of travel is clear and the net zero destination of power for data centres has been fixed. How quickly we get there depends purely on the pace of change and our own adaptability.