Australia’s government has said it now wants to abolish its Carbon Tax – legislation that would have made its largest emitters pay for carbon emissions. The move could stop such costs being passed on to customers. We asked our colleagues on LinkedIn - how will it affect overall design trends in the data center industry across the nation?


Tony Khoury, national business and technical director, JCS Technologies:
I really don't see it making any difference. It seems to me that the only way the data center industry in Australia is going to make the necessary changes to reduce power usage is through legislation because giving people the choice does not seem to be working. There are certainly some sectors of the market that are trying to make changes but for the most it is just not happening.

As an example, there was a recent public tender request for two new (smallish) data centers for a government department and their requirement in the tender request was for a power usage effectiveness (PUE) of 1.9. As a result they are going to get brand new data centers using air-cooled DX CRAC units and non-segregated hot and cold aisles for air flow because that is what all the respondents know will offer the least expensive day one tender price. There is nothing in the tender request documents to suggest that using less power is a strategy that the government wants to uphold and I am afraid that this is a sad indictment on the general market in Australia.

They should be seeking PUE's of less than 1.4 and in so doing they would also be defining the least expensive option they have when the life of the data center is considered.

There are many more examples that I could share but it is just unfortunate that this continues as I didn't see any real change even when the Carbon Tax was in place.

Chris Johnston, SVP and critical facilities chief engineer, Syska Hennessy Group:
A comment from a Yank. I find it encouraging that my Australian cousins have decided to eliminate their Carbon Tax and stop blindly following their misguided UK cousins. The UK version makes UK data centers less competitive in comparison with the rest of the world, and the revenues now go into their general revenue hopper.

There will never be a carbon or similar tax in China, India, Russia, US, etc., so why shoot yourselves in the foot? The energy efficiency of a data center should be determined by market and client demands, not politicians.

Hiroki N, manager for datacenter planning & administration, At Tokyo Corporation:
Legislation is not the only way to encourage lower PUE: Having electricity at 25+ cents/kWh will force data center users to look for PUE <1.5 purely because [the] monthly electricity bill will trump initial CapEx over five years. Looking at how oil and other fuel prices changed in the past five years, data center designers should make pessimistic assumptions about electricity prices in 2020, and then double it.

Tony Khoury:
I have tended to believe the same thing but it seems that this is not working. I don't believe that legislation is the right outcome but unless people make changes on their own we will eventually get the legislation we don't want.

There are so many new colo facilities being built where they pretend that they are designing for low PUEs, whereas in fact they are designing for low Capex. The only place they have low PUE's is in their marketing documents.

As another example, I was involved in a tender for a new defence data center in Australia a couple of years back where the customer would have saved approximately AU$20M over the life of the building (from memory I think this was 15 years) by implementing one of the new cooling systems and yet they still installed CRAC units and chillers. Lower CAPEX up front but much higher installed cost of ownership.

Chris Johnston:  
I agree. Astute owners want facilities designed for lowest TCO. But the lowest PUE does not always provide the lowest TCO. Also, when the cost of utilities is low, lowest TCO = lowest CAPEX.

Tony Khoury: 
I agree that when the cost of power and water is low then your comment is correct although it very much depends on the number of years that is considered for the TCO analysis. Most companies want payback in two to five years and so when the cost of power and water is low then this makes it difficult to justify.

In Australia, the cost of power has been rising dramatically for a number of years and availability of large volumes of water tends to change year by year. The cost of power over a five-year period in Australia (on average) must be assumed to be well over $0.20 (20 cents) per kW hr and at this rate the payback for low PUE designs seems to me to be a no brainer although this is not true of every site of course.

The real problem is that not enough customers are pushing for low PUE sites yet and secondly and possibly most importantly CAPEX is the domain of the colo but OPEX (cost of power) is the domain of the customer and when we ask a colo to consider higher CAPEX they sometimes don't see how they will get payback because they are not saving the cost of OPEX for themselves. I would have thought that the colos could charge more for low PUE space on a per m sq basis but I am not sure that this is correct.

Verghese Jacob, owner,  FMevolution (Australia and China):
I would like to join this debate by saying that while Australia enjoys some of the lowest pricing per kwh lowering the PUE at data centers is not just a green initiative, it is also a competitive advantage for data center providers. When you project the cost savings over a five, ten, 15 or 20-year period, the cost savings are significant and astute tenants get this.

Legislation does assist to drive behavioural change but year-on-year operational bottom line cost savings for tenants is a key factor in decision making,

Tony Khoury: 
That begs the question as to why so many continue to build new data centers without going for low PUE high efficiency designs? Who is stopping the change from occurring? Is it the developers, the consultants, the suppliers or just simply that the market is not pushing the colo market to do so yet?

Verghese Jacob: 
I am not sure but often it comes down to client discretion and budget available. All the newer and larger commercial co-location data centers we come across have design PUEs between 1.4 and 1.2, which is good. Smaller data centers design PUEs between 1.6 to 1.8 which is also good. Compared to where we were seven years ago with designed PUEs are high as 1.8 to 2.2, we have come a long way.

Hiroki N:
I can think of three possible reasons:
1. Lead time pressures: When IT departments are under strong pressure to deliver their services ASAP, they will head for the lowest lead-time solutions (ex. containers) regardless of TCO.
2. Perverse incentives: If IT department do not have to budget for their power consumption (as was the case pre-PUE even in the US), then electricity bills are "Somebody else's problem" as far as CIO is concerned, so they would choose lowest Capex solution every time.
3. C-suite Reality Disconnect: The C-suite still thinks its 1960s, when oil was few dollars a barrel and power cost was literally peanuts...