Move to a Mainframe, Earn Carbon Credits

BBCWatcher writes "As Slashdot reported previously, Congress is pushing the U.S. Environmental Protection Agency to develop energy efficiency measures for data centers, especially servers. But IBM is impatient: Computerworld notes IBM has signed up Neuwing Energy Ventures, a company trading in energy efficiency certificates, in a first for "green" computing. Now if your company consolidates, say, X86 servers onto an IBM mainframe on top of slashing about 85% off your electric bill each megawatt-hour saved earns one certificate. Then you can sell the certificates in emerging carbon trading markets. IBM's own consolidation project (collapsing 3,900 distributed servers onto 30 mainframes) will net certificates worth between $300K and $1M, depending on carbon's market price. Will ubiquitous carbon trading discourage energy-inefficient, distributed-style infrastructure in favor of highly virtualized and I/O-savvy environments, particularly mainframes?"

Comments

Full Circle?

[Aereus]

I do find it ironic that computing started out with large mainframes, and now it seems more and more likely that the majority of computing needs in the future will be met by terminals connected to mainframes via virtualization.

Re:Full Circle?

[asliarun]

and now it seems more and more likely that the majority of computing needs in the future will be met by terminals connected to mainframes via virtualization. That is indeed Big Irony.

Re:Full Circle?

[Ed Avis]

People had their own little sandbox in the old days too. If you were paying large sums for an account on a timesharing system, you'd want to be sure that some idiot wasn't chewing all your CPU time or memory. And you certainly wouldn't want other people having access to your files. Hence the elaborate systems to virtualize and isolate each instance, and quota out system resources fairly.

Please remember that in computing, nothing new has been invented since 1970.

Carbon credits = lame

[Z80xxc!]

The whole concept of "carbon neutral" and off-setting your carbon emissions for whatever reason seams kind of lame to me. Instead of continuing to do things that cause global warming while doing other things to supposedly reduce your "carbon footprint", why not just try to eliminate or reduce the problems in the first place? It's not just individuals, it's the whole mindset of society. Instead of going for carbon-neutral server farms, why not develop cleaner alternative electricity options to power those server farms? Solar power could do a lot, but we'd rather earn carbon certificates. It just doesn't make sense.

Re:Carbon credits = lame

[Antity-H]

Actually it does according to market theory.

Currently the market does not integrate the cost of emitting carbon in the atmosphere. As a result the carbon emitting technologies seem to be less expensive for the same result and the market logically develops these. Introducing a feedback in the market that the carbon emissions actually has a cost sends a message saying that carbon emitting tech is not the most efficient choice. The market will find an alternative solution instead of a solution being forced on it which might not be the most efficient in the end.

You mention that you want to eliminate the problem in the first place then you mention solar power, but how do you know that solar power is the best, or that nuclear power is? Maybe it's wind based, or ethanol based, or hydrogen based power or even cattle based power that's the most efficient. Or maybe a company will start doing research because there is a market for it and someone will come up with a transimentional p0rn energy extractor or even an Anonymous Coward based power source, who knows ?

The thing is the market will integrate the feedback signal and propagate it. This avoids forcing decisions on the market about the solution, the certificates are only reminding it of the problem. Going for carbon-netural server-farm is simply passing along the signal back to energy producers.

It looks like it's working for other problems.IIRC sulfur dioxid emission certificates led companies who claimed that installing an emission cleaner for it cost too muuch to actually install them even though buying the certificates seemed to cost less. the real price (vs company reported) of installing the cleaner was less than trading certificates in the long term thus they ended up investing.

Let's hope it will work for carbon too.

Re:Carbon credits = lame

[IamTheRealMike]

the IS NO COST to emitting carbon that anyone can quantify as yet

Sure there is. Just make an organization liable for the costs of climate-change related damage relative to the amount of CO2 it directly emits. You already have to buy carbon credits if you emit CO2 so we have a registry of who emits and how much. This way, the cost is amortized over the whole economy, increasing our ability to manage it (via general price increases).

For instance, let's say that mosquitos start moving up into Europe and spreading various nasty diseases. The health insurance claims for these events can be claimed back from the economy as a whole by suing the CO2 emitters in a class action suit. The details of whether a particular problem was caused or the risk increased by climate change can be thrashed out by the courts. I sense some scepticism in your remarks over whether climate change is real - that's OK, you can believe what you want, but I suspect when put in a court any such defence would have a hard time in the face of a nearly unlimited supply of expert witnesses. The CO2 emitters would be forced to try and calculate the risk to the environment from what they do based on what they believe and the advice their experts give them, and would then pass that on to their customers, thus the "true cost" of climate change would ripple through the economy.

This has benefit over the rather artificial carbon credits market, in that the "cost" of emitting a ton of CO2 is - as you rightly point out - basically pulled out of somebodies arse right now. What's more, they were deliberately set low enough to not have any impact on existing businesses, so instead of bringing about real change they just brought extra democracy. The idea of using markets to take action is the right one, but the "risk premium" needs to be priced into everyday goods.

I just made this scheme up off the top of my head. There are several key objections I can anticipate. The first is that climate change seems likely to kill a lot of people via disease/drought/etc, if indeed it's not doing so already, and how can you price a human life? Well, it is possible, but only in various untasteful ways. I don't think this one is solvable, nor should it detract from the scheme - the market is a tool and we need it to serve us now, to reach our end goals.

The second is that it would be inflationary if enacted globally, at once, because it would lead to a round of general price increases which would then in turn cause more borrowing by those without the spare cashflow to absorb it (ie, most people these days), thus inflating the money supply. This is especially true of essentials like oil (let's ignore peak oil for now). Inflation in the presence of a general price increase is not inevitable assuming you define inflation as an increase in the size of the money supply - that's an artifact of the fractional reserve. Replacing the fractional reserve with something less prone to inflation is certainly a good idea. But, if you suppress inflation (eg, by going to a Robertson/Huber type money supply), a general price increase makes us all poorer. That's more or less inevitable though - we would simply be paying what other people less able to pay (because they just lost their food supply/health/whatever) would be paying anyway, but everyone pays a small amount now instead of watching and saying "I hope that never happens to me". It's not a different concept to insurance in fact, but it's not optional, because climate change affects everyone.

The third is that it requires everybody to act more or less in concert. Unfortunately the "race to the bottom" is a general problem with regulating business and should not discourage us from working together to do so.

There are probably more problems with this scheme, but it does have the advantage that carbon emission is priced "naturally" and integrated into the sticker price of things like a unit of electricity - if you can get yourself out of the CO2 emitters game by replacing your electricity usage with solar or wind (or even nuclear!) then you are no longer liable for potentially huge disaster-relief costs, thus you can lower your prices, gaining an advantage over your competitors.

Re:Carbon credits = lame

[ErroneousBee]

My last remark in that comment was based on my immediate perception of the USA from Europe, sorry.

Hmmm... Let's list the first nation with an emission test for vehicles. (California 1966, USA 1968)
How about the first legislation on auto manufacturers for fuel efficiency (USA 1975)
Now, just to be sure, let's list the top five carbon emitting nations - per capita.

Qatar, Kuwait, UAE, Luxembourg, Trinidad and Tobago (weird)

I hope this helps to change your perception. Granted, some of our policies are misguided, or downright stupid, but that's a lot different than intentionally negligent.

Actually, lets list them all

And lets observe that the top 9 have a population of about 12million, and are all island, desert or city states.
Let us also observe that the major European states (UK, Germany, France, Spain) all have half the per-capita figures of the USA.

The reason the US eneacted those laws before Europe is because Europe was going for small and efficient anyway (E.g. by producing the Mini and VW beetle, and there was already pressures on fuel efficiency via fuel taxes and fuel rationing (during the war).

This attempt at spinning the figures, plus trying to shift the focus away from yourselves and small countries, most of whom are producing oil for the industrialised nations anyway, will only reinforce many perceptions about Americans.

Re:Carbon credits = lame

[StopKoolaidPoliticsT]

At the bottom it says "But overall, US adults have the biggest annual travel carbon footprint in the world at 7.8 tonnes, more than double France's 3.7 tonnes, which comes in at number two. Third on the list, at 3.1 tonnes, is Britain." -- the USA is a big jump ahead of France there!

For instance, "If one in 10 Americans used public transportation regularly, U.S. reliance on foreign oil could be cut by more than 40 percent--the amount we import from Saudi Arabia each year." (source). This notes that public transport use in America has now got back to the level it was at 50 years ago -- I don't know how much settlement density has changed in that time, maybe people have left cities a lot (?), but if it used to be possible, why isn't it possible any more?

I'd be surprised if 1 in ten Americans (30 million people) weren't using public transportation regularly. In NYC, it's a much larger hassle to have a car (fees, tolls, paying for parking, etc) than it is to take the subway. The same is true of nearly every city I've been to in the eastern half of the US (Seattle being the only city I've been to on the west coast and that was either walking or driving around with my girlfriend so I didn't check out their public transportation).

Outside of the cities (ie the vast majority of the US), public transportation just isn't possible. I live in a town of 7000 people with about 100 miles of road. We have a grocery store in town but I need to go 15 miles (one way) to get a decent selection of food at a reasonable price (a large national chain and another regional chain), to buy clothes, etc. I need to go 25 miles in another direction to get to specialty stores. It's not economically feasible to build a million little stores to service a couple dozen people each. It's also not feasible to expect people to walk up to 5 miles just to get to a public transportation station that will zip them off to those locations once an hour (damn, you missed it by 5 minutes and now have to wait almost an hour). Oh, double the buses, trains or whatever? Ok, you just doubled the carbon output (and costs) and halved the ridership of each transport. You certaintly can't solve the problem of distance to the station by running more transports around to pick people up (they'll be empty most of the time and will end up creating more carbon emissions than cars. For reference, it costs about $400k a year (not counting acquisition of new buses) to bus our kids to/from school on predetermined routes twice a day). The vast majority of the country (excluding dense, urban areas) is even far less dense than my town (190 people/mi^2 or 73.5/km2). I've been places where the distance to the nearest neighbor is measured in miles.

Finally, only half of the oil used by the United States goes to create gasoline. A quarter of it alone goes to home heating and most of the rest is used for farming and industrial purposes. If gasoline makes up 50% of our oil useage and 10% of Americans using public transportation means cutting oil consumption by 40%, that means that we'll see a 80% reduction in gasoline useage from a 10% increase in transportation. Why, at that rate, if we get 50% of people to use mass transportation, we'll not only stop using gasoline entirely, why we'll be creating 4 times more gasoline out of thin air than we consumed beforehand. Without looking at it, I'd say their numbers are flawed (or highly skewed to consider the 10% only those who use the most gasoline).

Why does the US use so much gas to get around? Again, it's a big place with a low density so mass transportation just can't work. People like having the freedom of getting in their car and going somewhere on their schedule rather than when some department of transportation decides they can move from point A to point B (and you're screwed if you miss the last ride of the night because you had to work late... it's not a matter of walking a mile pissing and moaning about your luck, you might be walking 50 miles or paying out the ass for a

A carbon credit is a permission to pollute

[Colin Smith]

At the moment there is no permission required to pollute, you could pump as much CO2 into the air as you like. Well, instead of that the government says:

1: You need permission to pollute.
2: You get those permissions from the carbon credit markets.
3: You have to buy them at whatever they cost in that market every year.
4: You can sell permissions if you have more than you need.

Then the government auctions enough credits to represent a slight reduction in the overall production in CO2. Each credit might represent one tonne of CO2. Then each year the government reduces the numbers of credits available in the market. The cost of a credit then increases simply due to the reduction in supply or the increase in demand.

As the cost of emitting the CO2 increases, companies will switch to alternative solutions, choosing whichever they like best.

Of course, this only works if politicians aren't completely corrupt or utter morons, as seems largely to be the case. In that case they might give companies credits and allow them to sell them on the markets, it's basically free money to those companies which receive the credits.

Power Consumption Predictions

[BBCWatcher]

As a general rule, if you're building a business computer and want to save as much as electricity as possible, the most highly virtualized (and virtualizable) platform wins. So attributes like massive caches and screaming I/O help enormously. (I think there was a Stanford study recently that figured this out.) Thus it's no surprise a modern mainframe is more energy efficient than anything else.

But in the Computerworld article referenced in the original story, IBM says the carbon program will also be available for its System p servers at some point in the future. My prediction is that you'll typically get fewer certificates if you move to System p versus System z, but it's likely businesses will do some of both depending on what sort of applications they're rehosting. There are some types of applications that will do better on System p, and there is some software that runs on AIX that doesn't run on z/OS or Linux.

Regarding SPARC it's impossible to say since Sun hasn't entered into any carbon credit auditing system yet. The IBM-Neuwing program is a first. However, my prediction is that you'll get even fewer certificates if you consolidate to SPARC. I say that simply because I assume IBM is acting in its own self-interest, and I'm sure they think the energy efficiency fight is one they can win against other vendors. In this case self-interest and environmentalism coincide. For any of these platforms, though, businesses will figure out whether the certificates favor certain platforms over others, and they'll do that application by application (or application function by application function). And many other factors will go into the decision as well, although most of those factors pull in the same direction as energy efficiency, such as software charges. One could even imagine that before long server vendors lagging in the energy efficiency department will start bundling carbon certificates with their servers in order to compete. Thus IBM adopting this program is a smart way to respond to an untapped market need and to raise the effective price of competing servers compared to IBM's. Very smart move.

By the way, the world has totally flipped on its head, and it would be extremely misleading to say an IBM mainframe is "proprietary" and X86 (for example) isn't. What does proprietary mean? You can run pure 100% GPL Linux on an IBM mainframe -- Debian, Slackware, CentOS, etc. -- and you don't even need a closed source driver as you usually need for X86 servers. IBM publishes extreme instruction-level detail in a free book called Principles of Operation, and it's so detailed and thorough that the open source community created an implementation of the instruction set called Hercules that actually works compared to still imperfect efforts like Bochs and QEMU. (Although IBM may assert patent claims on its processor architecture.) One company is porting OpenSolaris to System z, and they didn't even have to ring up IBM. In comparison, Intel and AMD also may assert patent claims, and AMD is suing Intel for alleged monopolistic behavior. Neither Intel nor AMD publish PoO-type documents (to that level of detail). Then there's Microsoft Windows, and it's hard to think of any more proprietary OS than that.

Also, IBM changed the way it charges for z/OS software about 7 years ago. Now almost everything is charged by the amount you actually use, something IBM calls Variable Workload License Charge (VWLC). If you run a little bit of DB2 in one LPAR (partition) but a lot of IMS in another, then you pay a little for DB2 and more for IMS. You also control exactly what you consume using something called softcaps, and you can set those either per-LPAR or for a group of LPARs. One interesting little twist to mainframe subcapacity licensing is that, if you need a little bit of WebSphere (and a lot of other IBM products), the lowest entry price (smallest license you can order) is for z/OS. You can order as little as 1 "Value Uni