A couple of years ago I reported on the plans of the Federal and State governments in Australia to develop new metrics to measure energy efficiency in data centres. The idea was to add specific metrics for data centres to the existing National Australian Built Environment Rating System (NABERS) - similar to the Leadership in Energy and Environmental Design (LEED) rating in the US.
The task is now complete and the NABERS Energy for data centres rating system claims to be the first in the world able to measure and score the operational energy efficiency of IT equipment housed by a data centre. There are now three NABERS rating tools available for data centres owners, operators and tenants:
IT Equipment – measuring and benchmarking the energy efficiency of the IT processing and storage capacity within a data centre.
Infrastructure – uses the Power Usage Effectiveness (PUE) to assess the efficiency of the building services (e.g. air conditioning and lighting). There will be an independently assessed and accredited performance rating available for data centre operators.
Whole Facility – combines the IT Equipment and Infrastructure rating metrics to assess the energy efficiency of the data centre as a whole.
The rating system will be managed by NSW OEH, the National Administrator for NABERS rating systems for buildings.
Review: As I said when the rating system was first mooted, I’m not keen on the proliferation of standards around green IT – the more universal measures are then the more likely they are to be adopted, particularly internationally. But the most widely accepted metric – PUE – whilst very useful, is limited in the long term.
In my last post I mentioned eBay’s initiative to go a step further and this Australian effort also builds on the PUE. The problem is that they come from different directions. The NABERS methodology is a buildings approach, so focused on energy efficiency of the facility as a whole. eBay is looking more towards maximising the energy efficiency of individual customer transactions, i.e. the IT workload.
Both are valid. But it raises the question as to whether it’s actually possible to have one universal measure of energy efficiency in a data centre, or does it depend on who’s asking the question?