Power Usage Effectiveness (PUE) is well understood – it’s been a useful metric for determining energy efficiency since it was created by The Green Grid (TGG) in 2007 and adopted by the ISO as a KPI for data centres.

The manner in which compute is cooled is an enormous contributor to PUE; and, since the dominant cooling technology for data centres has been chilled air, this measurement assumes air cooling as part of the equation.

With sustainability and carbon footprint considerations driving modern data centre design decisions, liquid immersion cooling has emerged as the pre-eminent way to conserve energy while eliminating heat. Yet PUE is based on the old norms of data centre architecture and design, and isn’t the best metric moving forward. 

Data centre owners and operators use PUE to calculate energy use in liquid cooled facilities because it is relatable and familiar. As more data centres migrate to liquid cooling to improve efficiencies and move toward more sustainable operations, it is time to rethink how we measure performance.

We know that moving from chilled air to immersion is a fast way to gain PUE efficiency. But does it provide a useful measurement for comparing liquid cooled data centres to each other? In fact, we need other metrics to get a more granular and more accurate picture of how all cooling technologies impact power use and to enable better decisions.

This is a transformational moment for data centre design, efficiency and sustainability. PUE is the classic case of trying to fit a square peg into a round hole. And it is time to think differently about how we measure efficiency. There is radical change in how data centre operators reduce the temperature of their valuable computing equipment, to feasibly enable them to balance the competing needs of sustainability, regulatory compliance, costs, capacity issues, and computing power. Liquid immersion cooling is necessary for this type of change, and the old tools no longer apply.

Beyond PUE

With so many new technologies changing the essence of the data centre, it is crucial to find the most useful metrics to enable data centre owners and operators to make decisions. PUE is a blunt tool, measuring overall power use but not looking at the impact of individual contributors nor weighting the value of the type of cooling solution – chilled air versus liquid. 

Today, data centres are necessarily focused beyond simple energy efficiency. To remain competitive, data centres are also concentrating on decarbonisation and the use of renewable energy sources. So, if PUE is obsolete, what are the popular metrics that are lining up to replace it?

Carbon Usage Effectiveness

CUE is the ratio of the total CO2 emissions caused by total data centre energy consumption to the energy consumption of IT equipment. It was developed by TGG to measure data centre sustainability in terms of carbon emissions. Liquid immersion cooling reduces energy consumption by eliminating moving components, such as fans.

Cooling Capacity Factor

CCF is the ratio of total running rated cooling capacity to 110% of the critical load. It was created by Upsite Technologies to benchmark rated cooling capacity versus utilisation; to determine and understand current inefficiencies in your cooling infrastructure, as well as identifying opportunities for improvement.

Energy Reuse Effectiveness/Energy Reuse Factor

Energy reuse effectiveness (ERE) measures the overall energy efficiency if energy is being reused outside the data centre. Energy reuse factor (ERF) measures how much energy in the data centre is reused elsewhere. These metrics, which are two sides of the same coin, were developed by the TGG to capture use of excess energy, which cannot properly be measured by PUE. 

Including use of waste heat in a PUE measurement could return a PUE of less than one, making it difficult to effectively compare metrics across data centres.

This is a transformational moment for data centre design, as efficiency and sustainability grow in importance. Properly measuring data centre efficiency, and doing so in a way to enable an effective comparison of data centre cooling technologies, is a complex problem. 

PUE was a foundational metric which enabled data centres to build awareness of their own energy usage. But it is a blunt tool and one that has outlived its usefulness – it is inclusive of the entire data centre and cannot allow fair and direct comparison of the contributions of individual elements of the data centre infrastructure.