Posted by: mikepearsonnz | September 7, 2009

The idea of the “NZ Inc.” Datacentre

Datacentre

Datacentre (by) Erwin

Laurence Millar mentioned the idea of the NZ Inc. datacentre at a recent NZ Computer Society event.  He gave me a HT for the original idea.

Background

I raised the idea of a NZ Inc datacentre in 2008, after a visit to the US.  That visit got me thinking:

“How will New Zealand (government and business) remain globally competitive, in a cloud computing environment, when the US can compute for 5 cents per hour?”

It led me to question the way we do our computing in New Zealand – it’s not to say we do things wrong – we do things differently because of a variety of social, political and environmental reasons.

As one example, with the current price of NZ electricity alone (~ 20 cents per kilowatt hour) are we likely to achieve a competitive level of computing?

Services that can, will eventually move offshore, driven by price.    Not all services will be able to move, because computing is not strictly a commodity like electricity.  Computers process information, and information has different values.  Some information will make more sense to process information in New Zealand.  The State Services Commission has a discussion paper on this topic: Government Use of Offshore ICT service providers.

For information processed in New Zealand, it will still need to be processed as competitively as possible.   The competition is global.  Otherwise inefficient NZ computing will be an additional overhead/barrier/tax on every associated NZ good and service.

__________________________________________________________________________

NZ Inc. datacentre

So that got me thinking – how do we create an environment that encourages innovation and competition that ensures NZ datacentres are as competitive as possible?

My idea was to create a “NZ Inc. datacentre”.  It would be the host for 4 datacentre providers.

  • Each provider will have the same fixed physical input (power) – no additional power will be provided.   This encourages doing more for less, with providers seeking to achieve better Power Usage Effectiveness (PUE) scores.
  • Each provider will be penalised for waste outputs, such as heat or e-junk.
  • The first provider to develop a proven innovation, will be entitled to royalties from the others; to encourage sharing of the ideas.
  • Every 5 years, the place of the worst performing provider will be open to challenge by other potential providers.

__________________________________________________________________________

Requirements for the NZ Inc. datacentre?

The requirements for a efficient NZ Inc datacentre would not be trivial.

  • Access to 400 megawatts power – this is a working number based on this article x 4.  As a comparison, Tiwai Point aluminium smelter uses about 540 megawatts per year.
  • Access to cheap, clean, renewable power – the most likely source in New Zealand is a hydro electric dam, which can also provide water for cooling.
  • Cool outside temperatures – which can be used to cheaply cool the datacentre.
  • Land near the power source – the NZ Inc. datacentre should be as close to its energy source as possible, to avoid electricity transmission loss (There is no single figure I can find, but the loss of transmitting power on the national grid is suggested at slightly less than 10%).
  • Redundant fibre to New Zealand – to transport the data.

The logical place for a NZ Inc. datacentre is probably somewhere in the South Island, like Benmore.   It’s not going to employ many people directly, but has the potential to assist NZ  Inc. in competing with rest of the world.

__________________________________________________________________________

Is there a likely international  market for NZ Inc. datacentre services?

When I’ve mooted the idea, people have often asked if there is an export market for the NZ Inc. datacentre.

Exporting “computing resource” is an attractive idea, because it has some unique characteristics:

  • 1 unit of power can produce variable value – 1 unit of power in the physical world typically produces the same outcome every time (whether it be grass->lamb; or electricity->aluminium)
  • It is cheaper to move photons, than it is to move atoms (physical products) or electrons (electricity).

The NZ Inc. datacentre will face challenges in selling capacity offshore, for the following reasons:

  • Communication Link Redundancy:  A representative of one international company that builds such datacentres said to me “I don’t even consider countries with less than 3 high-speed pipes to the world — preferably I’d like 5 links, with maybe 1 or 2 satellite earth stations”.  New Zealand currently has 1 link, and is struggling to justify a 2nd.
  • Latency: Latency matters for certain applications.  Google demonstrated that a delay of less than half a second on their search results page, had a measurable impact on the number of searches per user.  A daily impact of 0.5% is of real consequence at the scale of most global Internet sites.  Therefore datacentres will be hosted near large population centres.
  • Scale: In the future, even a 400Mw datacentre will probably not be “large” by global standards.  We will reach a point where NZ cannot fund capital investment to the level to compete.

Perhaps you can think of some competitve advantage that a NZ Inc. datacentre could offer the world?

__________________________________________________________________________

Chances for a NZ Inc. datacentre?

The chances of a NZ Inc. datacentre are low.  NZTE and the SSC had an exploratory meeting with members of NZ industry in 2008.  The general feedback was individual NZ companies will meet this challenge and there was no desire to develop a nationwide initiative.

What do you think?

Advertisements

Responses

  1. If you look at Porter’s strategic thoughts, having an NZ data centre would not be helpful. How would we answer these questions:
    1. The threat of entry of new competitors
    2. The bargaining power of suppliers
    3. The bargaining power of customers (buyers)
    4. The threat of substitute products or services
    5. The rivalry among existing firms in the industry

    In terms of competition, countries such as ices India, China, Singapore, Philippines, Malaysia, Vietnam, US and now upcoming South America can offer the closest of challenges to what NZ inc Data center can offer. Like you said, the biggest challenge here would be utility power unless a price negotiation can be met or they themselves set a NZ inc Data center. Then again, what about resources? From my experience, resoources are cheap coming from Asia and even its from the US, these US firms then sub-outsource internally or externally for a cheaper price than what we can offer. Should we compete ourselves in terms of price then we would have to change the labor laws in NZ to meet those challenges.

    Price cannot be one of the determining factor to compete with other global Data Center incs.

    However from my experience, we should tap into our own expertise for a Data center that focuses Multi-media technology (thanks to Peter Jackson and his cohorts for opening up NZ’s expertise in the global multi-media market) We’ve got a couple of NZ firms that have scored big time in the entertainment industry (I know of 2).

    My thoughts on the matter on a boring Saturday weekend

    • Hi, thanks for your comment.

      I agree that we cannot compete on price, to draw customers to a NZ Inc. Datacentre. The question is more like “How do we make NZ datacentres competitive?”. If NZ customers must use a NZ datacentre, and that datacentre costs a lot, then that cost goes on every item exported from New Zealand – a financial drag.

      It is not just NZ pricing of datacentres. It is also NZ pricing of international bandwidth. Traffic costs will penalise multimedia industry in this country. Rod Drury did a very good opinion on this, in the NZ Herald this week, http://www.nzherald.co.nz/business/news/article.cfm?c_id=3&objectid=10597700&pnum=0.

  2. Sorry for the typo errors! I have big fingers!

  3. Hi Mike,

    Nice topic of discussion.

    A datacentre is a major investment for companies and unless there is a significant coalition representing a growing need for this, it’s not going to happen.

    Datacom Systems has recently completed their datacentre at Albany. It has the capacity to accommodate 644 modern high-density racks and has the flexibility to accommodate any type of rack or equipment. It took more than 100,000 hours, more than 250 people and 120 sub-contractors to build.

    Most of the Datacom customers have now been relocated to it and it’s been running like a dream. Off course there is site resiliency to their other datacentre around NZ.

    The final cost ended up being $30M, 5M up from the $25 scoped.

    The PUE rating is 1.4x which is the best in NZ (above average) and close to being state-of-the-art.

    While not boasting, the point I’m trying to make is that this is a significant investment. Without proper planning and lack of drivers, the project would have failed miserably. So unless there is a need out there, the requirement for “NZ Inc.” is minimal. Considering the rack capacity Datacom’s “Orbit” datacentre has, and considering the move towards virtualisation for most of the IT industry, “NZ Inc” investment may not break even for a while.

    Following for your interest:

    http://www.stuff.co.nz/technology/563447

    http://datacom.co.nz/About-Datacom/News/PM-Salutes-New-Data-Centre.aspx

    http://computerworld.co.nz/news.nsf/news/3B87A702386A51C8CC2575B0001F87D6?Opendocument&HighLight=2,datacom

    Best regards.

  4. Hi Imran,

    Datacom needs to be more widely congratulated on its datacentre design achieving a PUE of 1.4 today. The EPA estimated that in 2011, a “normal” datacentre would have a PUE of 1.9, and a “best practices” datacentre would be 1.3.

    How is Datacom working to reduce that PUE figure further?? Google is averaging 1.19 and one of their datacentres recently achieved 1.11

    I agree that datacentres are a major investment and it’s often a chicken/egg scenario, you need guaranteed customers before you can invest.

    Perhaps the NZ Inc. datacentre idea I propose would help with this, as it incentivises greater efficiency, and removes the need for capital investment in land, power.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Categories

%d bloggers like this: