blog How much impact will the Federal Government’s so-called Carbon Tax have on server hosting costs? According to Aidan Tudehope, the managing director of Macquarie Telecom’s hosting division, quite a lot. Tudehope has published a decent analysis of the situation on the blog of Macquarie Telecom-associated cloud computing provider Ninefold. A useful excerpt:
“Lets select one of the most common servers available today, HP’s Proliant DL380 G7 (with say 2 Quad Core; 2 DDR-3; 2 HHD; 2 Power Supplies). Simple enough. This server consumes 256 watts, which at 16c/kwh costs approximately 4c/hr to run. So far this doesn’t sound much but these few cents add up to over $350 every year.
But don’t forget that servers create heat (that’s why they have fans (and sometimes big ones) at the rear) which needs to be cooled. If you are housing your servers in your office, then you will be using at least the same amount of power to cool the servers as it takes to run them. In many case even more. So we now add another $350 to cool the server – ouch!
Add a little for the new Carbon Tax, say 12% or $84 pa. We are now at $784 just for the power to run (and cool) a server for a year. But no one has 1 server, well not the last time I checked. In fact you can pack 42 servers into a compute rack and draw 10KW, and for that privilege, you’ll be writing a cheque to your local power company for over $32,000 every year.
Now that’s real money, and a real carbon footprint.”
It seems reasonably clear that the Carbon Tax will have an impact on Australian server hosting costs, and of course datacentre costs in general. Consequently, I’m sure chief information officers and IT managers all around the nation are currently calculating that cost and discussing it with their chief financial officers. Not fun; but of course, there is an argument to be made that there is a great deal of long-term value in environmental initiatives such as this; and everyone has to do their part. Just another rationale for rationalising your server footprint a little further, perhaps ;)
Image credit: Whrelf Siemens, royalty free
16c p/kwr WTF? Where does this guy get his power from :)
Yeah, it’s more likely to be 29c per kWh!
Any company with any competent CIO & IT Management will have their Equipment in a Data Center so these costs are not really a true reflection of reality.
The current state of the art Tier3 DC’s in Aus from Adam Internet, HP & NextDC have a PUE rating between 1.3 and 1.6. Using 1.6 PUE as a worst case scenario, this means for every 1000watts of power being used by the servers it requires an extra 600watts of power to cool them and provide them with network connectivity, redundancy etc. In a DC with a PUE of 1.3 it’s only an extra 300watts of Power per 1000watts of servers.
That’s pretty darn efficient and nothing like “double the cost” to cool your server!
Having your servers in a proper Tier3 DC (even if you are a small Biz) provides substantial de-risking benefits to your business due to redundant power, redundant and geographically diverse Network provisions (also most are carrier neutral so you can have carrier diversity too), redundant air-conditioning and so on.
10KW? meh, you should see what a DSLAM pulls, easily over double that.
Wasn’t that just a long way around saying that the Carbon Tax will add around 12% to power costs, whether for servers or not? What it costs to run 42 servers is irrelevant, what is relevant is what it cost before and what it costs now.
So whats happened every other year when electricity has gone up 10-15%?
These sort of price increases with power have been happening for years, why is it suddenly an issue now? All this story does is perpetuate the current myth that every price increase is because of the Carbon Tax.
There are plenty of other reasons prices go up, ranging from inflation and CPI through to good old fashioned greed.
The important here is to remember that the carbon price is exactly that. A pricing mechanism.
The CEOs and CIOs deciding that the increased costs need to be reduced, should be looking at using less grid-generated power. Add some solar power, or wind power, or whatever.
The price increases are deliberate, to encourage people to “find another way”.
Don’t want to pay the carbon price, use less of the service that adds that price to your inputs.
Exactly as it’s designed to do.
I agree, but really the effect on electricity prices in this regard is a secondary effect. The primary effect is to price carbon on the 500 biggest polluters, which electricity generation falls under. As more and more of our electricity is generated by non-carbon sources, and those methods of generation get cheaper, the secondary effect on the use of electricity will lessen.
Certainly.
But as electricity generators lessen their costs through employing greener technologies, and therefore lessening their carbon price costs, can you honestly see them dropping their retail prices? What business wants to drop their prices?
No – the prices will stay (relatively) the same – (indexing roughly against inflation) – so their customers will be paying relatively the same.
It’s costing the generators less to produce, which increases their profit margins. They might be faced with upfront costs to get there, but in the long term, it’s beneficial financially for them.
And they are producing less carbon.
If the customers are also using self-generated power, carbon is reduced further.
Here’s an example, based partly on fact, and partly on fantasy.
Not too long ago, I saw a story about fishmongers. They were complaining that to refill their fridges with coolant, the price had gone up from $4,500 to $20,000. Comment was that the price had gone up $285 per kg of coolant. Thats the fact part, assuming the numbers are correct.
So, they pay $285 more per kg. For starters, at $23/tonne that means the process is generating 12 TONNES of emissions to make 1 KILOGRAM of product… Yet they cant see a problem?
Anyhow, on to the example. In the past, there were two options. 1 costing $100 per kg, other costing $250 per kg (the fantasy part). Obviously the cheaper alternative is the more popular. Fast forward to post-CT world, that $100/kg option is now $385/kg, while the previously more expensive option is still $250/kg.
As an industry, they collectively move to that $250/kg option, because its now the cheaper option. The $250/kg option doesnt pay carbon tax, and because its now the preferred option, can afford to work on ways to provide the product cheaper in the long term.
End result is a cleaner product, with development within that product line to make it more efficient. Winners (mostly) all round.
And thats what has to happen with electricity as well. Move from the emitting option to the non-emitting option. As a business, you might invest in solar panelling on the roof for example, and have that investment pay for itself over time.
GongGav: the price rose so much because most refrigerants are hundreds or even thousands of times more effective as greenhouse gases than CO2 is. Luckily we only emit relatively small amounts of them…
Yeah, I know that. I was more looking that to create 1 kg of product meant 12 tonnes of byproduct if the $285 increase is fully because of CT…. There arent many processes where thats an efficient practice, to say the least, unless your talking precious metals or gems.
That 1 kg of product may be the most effective coolant, but if there is a product thats 80% as effective, but significantly cheaper, that will end up being the most used product.
Which is where Michael is coming from – its cheaper now because its not being hit, so businesses will adapt to use that cheaper option. Cheaper doesnt mean more effective.
Acutall, the ACCC has already come down on the refrigerant business for missleading claims.
They actuall do other things than just “rubber stamps” NBN deals.
http://www.accc.gov.au/content/index.phtml/itemId/1067053/fromItemId/142
[Quote]
ACCC accepts undertaking from refrigeration contractor for misleading carbon price claims
The Australian Competition and Consumer Commission has accepted a court enforceable undertaking from South Australian refrigeration contractor Equipserve Solutions Pty Ltd.
The undertaking relates to statements made by Equipserve Solutions in an email to its customers which attributed the entire amount of an increase in the price of a refrigerant gas to the carbon price when that was not the case.
“The ACCC was concerned that Equipserve Solutions’ statements had misled customers and caused them to believe that the entire price increase was due to the carbon price – when that was not in fact the case,” ACCC chairman Rod Sims said.
[End Quote]
Michael,
They might lessen their *carbon* costs, but low- or non-carbon sources of electricity are usually more expensive than coal without a carbon price.
This is because the long-term effects of carbon pollution were, prior to 1 July 2012, completely unaccounted for in electricity generation in Australia. Even now, $23/tonne is pretty cheap. The economic analyses I’ve seen price these external costs at between $65 and $800 per tonne, depending on your discount rate (i.e. whether you care about what happens to your great grandchildren or not).
Personally, I think the real cost is higher than that. How much is it worth paying now, to avoid the effects of 60-70 metres of sea level rise (assuming the long-term projects of those pesky climate scientists are correct). How much will it cost to relocate every major city in Australia, other than Canberra, to higher ground? What impact would that have on the Australian economy? What about some of the other projections, for swings between blistering drought and incredible flooding rains, for heat waves that *require* you to have air-conditioning if you don’t want to die of heat exhaustion? None of these impacts are likely to occur for centuries (we hope), but what’s it worth paying now to avoid a potential complete collapse of civilisation 300 years in the future?
Back on-topic – yes, the carbon price will increase the cost of running data centres. That’s the point.
But as others have already pointed out – the impact of the carbon price has already been overwhelmed by the increases in electricity prices over the last few years. Cost of electricity here in Qld has more than doubled over the past decade or so, and is predicted to rise another 30-50% in the next few years, so the carbon price impact is really only a very minor part of the story.
Great comments here. Pretty much already said what needs to be; it’s not JUST the Carbon Tax that increases server cost and it is, in fact, a little disingenuous to say it plays a large part when overall efficiency of your Datacentre would contribute more overall.
Also, 12% for Carbon Tax added?…..sorry, but IPART just allowed 9% increase from the Carbon Tax…..12% is a bit misleading, because that other 3% electricity generators have added “for other reasons” including infrastructure and “because they can”…..
Hardware has been becoming more and more power efficient over the years, which is good.
Problem is, as hardware gets more efficient, software developers get lazy, and dont bother to try and minimize hardware requirements for their software. But of course, some increased demand is expected as software adds more and more functionality.
Things like choice of language effect efficiency of software, and hiring developers good with algorithms instead of good at churning out cheap code.
10 year old server software would probably run from a mobile phone, with those reduced power requirments.
1000W? – time to buy the latest 12th Gen servers from Dell. Check out the video http://youtu.be/FkPrPaFt1wY
So this story isn’t telling us what Macquarie Telecom is doing to reduce its carbon footprint.
Better-designed server storage, for better air circulation and less need for cooling?
Lower-power chipsets, and power-managing chipsets?
Plant a few trees on the roof, or maybe put up some solar panels?
Actually, that all sounds way too hard – let’s just complain about a new impost that’s intended to change our behaviour.
How does Macquarie Telecom currently account for its carbon usage? Oh, it’s not considered a resource? Well, maybe it’s time to change that thinking, and think about what your impact is on your environment. It’s not just about money, it’s about the world you live and operate in – and companies just don’t want to understand that.
The way the climate is heading it is going to cost more to cool datacentres. Now might be an opportune time to save money and innovate.
Pfft whats $1,000 a year hate to be the barer of bad news.. but $1,000 / $10,000 / $50,000 wont make a difference on our federal budgets…
It costs us $300 to have a DNS record changed.. (yep simple a-record) the same price to run a server for a year…
CT will make no difference what so ever..
And shall we also point out that with cloud computing / virtualisation agency’s / the TCO has actually decreased.. (we went from 16 hosted server racks down to 9)
Sorry Renai you are better off focusing your articles on issues / costs that you actually understand
That’s not fair on Renai, Michael. This was an editorial opinion piece primarily by another author. Renai was simply reposting it on his behalf as an analysis/opinion point of interest.
If you don’t agree with it, fine, but it’s a bit unfair to say Renai “doesn’t know what he’s talking about.” He’s worked for ISP’s before. I think he has some idea….
Sorry, but Michael’s comment is correct! and he made a fair call! ( I didn’t take it as being nasty)
The article clearly states:
“Blog, Enterprise IT – Written by Renai LeMay on Friday, July 27, 2012 12:16”
Glad to see Ranai talk about the issue, as this will no doubt come to haunt those who haven’t already jumped on the Visualization bandwagon!
Nice post Renai. Ultimately what matters is how efficiently you can calculate power consumption to your hardware.
Refer to the following article to effectively calculate power usage for HP servers/ storage.
http://virtualcloudzz.blogspot.com.au/2012/12/hp-blade-powersizer-tool-explained.html
Comments are closed.