GreenITStrategy.com

So even though I am a Green IT freak I don't generally talk much about data centers - the main reason is because my background isn't in data centers - it is true I am an infrastructure person (not a developer) but I generally prefer to work on the business interpretation & high level systems architecture side of things than get down and dirty with server room designs and cooling efficiencies.  This kind of makes me something of a traitor in some Green IT circles :)

So rather than talk un-credibly about data center design, I thought I'd point you to a great post by Mike Manos, head honcho for Microsoft's worldwide data centers, who has announced a very detailed plan of our Generation 4 data centers.  The "eco" bits to watch out for in this post include:

• Non-critical/geo redundant hardware needing lower (or no) infrastructure requirements i.e. cooling, UPS, chillers
• Achievement of excellent PUE (power usage effectiveness) levels
• Elimination of water

One of the other areas that Mike mentions in his post is the reduced environmental cost due to not having to build with concrete - Mike asks in the post how much carbon goes into making concrete?  And the answer is about 5% of the world's total carbon emissions come from concrete!  It mainly comes from portland cement (one of the major components of concrete which is produced through a highly energy intensive process) however there are companies innovating in this space to replace portland cement with a polymer and normal aggregate from other industries that consider it waste (e.g. slag from the iron industry) - if you're interested in this, check out e-crete.  Gosh aren't I a little font of irrelevant knowledge.

There are also two other good links to visit while you trek the data center hike:

• A good video posted to Soapbox which provides a simulation of gen 4 data center concept.
• Christian Belady, power & cooling architect, and David Gauthier, data center architect, from Microsoft also posted a followup on Mike's announcement, outlining the infrastructure components and efficiencies from the plan Mike announced.

There is a hidden environmental impact that may not be immediately obvious from this data center design.  Its one of the elements in a cradle to cradle scenario:  when you account for carbon (or any other environmental element), you (try to) account for the entire lifecycle of a product (EPEAT is one standard which helps from an IT perspective to show you the impact of certified vendor's hardware overall lifecycle).

Back on the data centers, from a carbon perspective, the shipping of bulk loads versus shipping many small racks or individual servers is something that sometimes isn't thought about.  However considering transport is one of the largest generators of carbon emissions (energy of course being the other major generator), you can reduce your own data center's carbon impact by buying and shipping in bulk - even if its not at the lofty levels of Microsoft containers.

I was at the Green Business conference in Melbourne last week and one of the freight companies, Linfox, talked about the fact that their trucks run empty about 30% of the time.  This means their trucks are generating 30% extra carbon for no good reason.  As such if IT could more efficiently plan and utilise the freighting schedules and ship in bulk rather than individual lots, the overall supply chain carbon will be reduced.

There is the question of course of where the data center components are being shipped from and to.  One element here is asking suppliers to provide their own carbon impact statement for the hardware that you purchase for your data center.  And that is probably no where near as easy as it sounds for a supplier - one good example is HP:

• HP ship over 1 million products a day
• Most of the assembly is performed in Asia while the largest proportion of sales are in Europe & America
• They estimate over 2million tonnes of CO2e are generated annually from the supply chain, most of which are from international air freight and 25% from road transport/parcel freight (1/2 million tonnes).  They aren't certain because there are so many suppliers they can't possibly drill down level after level of suppliers to gain an accurate figure.

However we will see an increase in the future in the requirement to measure the environmental impact of the entire supply chain, not just our own backyards.  From my perspective this can't be achieved without partnership and collaboration of the involved parties.

As an example, what could be achieved if a hardware vendor and their shipping company enabled you to view their scheduled shipments, automatically correlate truck capacity to your order, and allow you to schedule your equipment to be delivered by both the lowest carbon method and at the lowest possible carbon impact (e.g. in a truck that normally would be empty)?  Think of it like being able to pick your seat on an airplane flight - if there is capacity and that seat is free you can have it.  Then add the element of automatic carbon calculation based on various factors like distance and shipping method (e.g. rail, truck), and put it in a nice GUI where you are able to easily pick and compare and you have a system that enables greater end-user control of carbon management from the supply chain (and not just for IT purposes).

I can't wait until some of these really intelligent, federated systems come into place in the future...yeah I've told you a million times before: I am geek...and totally proud of it.

References:

HP's Stats on Supply Chain carbon: http://www.hp.com/hpinfo/globalcitizenship/gcreport/energy/logistics.html