The SCSI Trade Association website has run an article entitled The Data Center’s Green Direction is a Dead End by Steve Denegri. It argues that the storage industry is effectively in denial and that we need more energy, not less, for the health of our industry. Denegri is a storage consultant and financial analyst. His pitch is …
Back to the caves
The Greenies won't be satisfied until they've destroyed all industry, not just data storage.
Then we can sit around trying to fend off malaria with organic, homeopathic "medicine", while we contemplate the wickedness of our ways.
Praise be to Gaia!
This guy is actually right, but ...
... the effect extends to the whole IT industry, not just storage. The whole information technology industry is inherently energy-hungry.
File under WTF
"the automotive industry saw a major transition to energy-efficient products beginning in the late 1970s".
Do they now all make bikes or what?
Anyway, bollocks. Adapt or die. If storage companies can't find ways to add value to their services beyond gigantism then they can build their own fucking power stations for their Hummer-branded data stores. I think organisations would find that by managing their data better they wouldn't really need the terabytes of multiple redundant email backups of kitten videos sent around accounts.
Work smarter, not harder and that.
whilst I agree with the gist of what this guy is saying, there is a finite amount of energy available to any industry, and the human race in general (at least until fusion reactors are realistic, if politics ever allows them off the ground due to tumbling energy prices, hence tumbling tax income).
As such, all industries should be working toward a more efficient use of said energy, getting more for less before the oil runs out.
Don't just cut energy use
While part of the basic premise may be right in that energy saving can only go so far, there is a lot more that can and should be done. The article mentions virtualisation, yet misses the fact that a smaller number of servers also means less energy is required to make, deliver and ultimately dispose of them.
There is also the point that many organisations spend – needlessly - on storage, because they don't manage their data effectively. Instead of archiving etc., it is viewed as easier to just buy more storage, which requires more power and cooling.
"Assuming a perfectly spherical cow..."
Sure, we'd all like unlimited free energy (maybe not Centrica or BP, but you get the idea). Hey, I'd like world peace and a pony. But if this guy reckons that the storage industry will be fine as long as they rewrite most of economics (and, most likely, the laws of physics) then someone should take him to one side and suggest that he go and peddle his ideas with the rest of the nutters on alt.kook.
"The cold, hard truth is that an ample supply of energy is necessary to grow any business over the long-term,... the harsh reality [is] that a sufficient amount of energy is, unfortunately, not available to keep the industry growing."
Can't argue with that. Similarly, the football industry has to handle the harsh reality that there's only one Pele.
But here's the answer: we need to clone Pe... sorry, to "expand the capacity of the power grid" instead of "embracing the energy efficiency paradigm".
Fiat lux indeed.
So most of you seem to be saying that the storage industry should fix it's problems by telling it's customers to use fewer of their products. That'll go over well with the stock holders, I'm sure.
Most of the point of this article, I think, was that R&D spent on making storage more efficient is increasingly worthless, since it'll inevitably cost far more than any potential savings in energy. And trying to do more with the same amount of energy consumption will limit future growth, which this guy claims is what kills industries.
That 70's US automaker comparison, while I don't really understand why it's relevant, refers to the fact that there were loads (30+ I'd estimate) of US auto manufacturers before the 70s oil issues / the rise of the efficient Japanese auto, and afterwards there were just 3. I don't have any idea why the gas prices lead to fewer manufacturers, except that maybe the total collapse of profit left only the biggest players alive.
A few of you don't understand the difference between necessary energy use and wasteful energy consumption. A Hummer (the civilian models) uses energy wastefully, since it doesn't do anything that a station wagon can't, except theoretically off road, which nobody actually does anyway.
But everyone likes Aluminum, which happens to require a huge amount of energy to refine. There is no way to make aluminum from bauxite without expending something like 10 times as much energy as it takes to refine steel from it's ores. But you don't see hippies telling us we should stop using Aluminum.
In the same way, it is necessary to use a certain amount of energy to get a certain amount of storage, and if someone needs a lot of storage, they need a lot of power, no matter how many 2% efficiency gains the storage vendors make. Because they can't make that many.
And if power capacity doesn't expand, then obviously the entire IT industry will have a hard time selling kit, since none of their customers will have anywhere to plug it in.
A Few Additional Thoughts from the Author
Hi, all. I wanted to quickly thank The Register for mentioning my article. I've been a big fan of this web site for a number of years.
I just wanted to respond to a few items in the comments and show my appreciation for your opinions.
In no certain order:
1. "Contrarian Thinking with a Vengeance": This article does go against consensus, but I want all of you to know that these are my true concerns with not only storage but data center computing, as a whole. In full disclosure, I composed this article free of charge for the SCSI Trade Association (STA) newsletter, so I don't have a score to settle. Nor am I trying to create alarm for the sake of my wallet. The insistent tone was meant to be a wake-up call for the storage industry for which I have a passion. Unfortunately, I believe the industry is going down the wrong path. Since storage providers are the audience of the original article, my concerns are focused in this category, but I certainly believe that this is a problem that hits enterprise computing as a whole and not merely storage. I feel as though the lack of sufficient electricity will hit the deployment of servers, first. The storage industry will feel the economic pain, as a consequence.
2. "Finite Amount of Power": This is probably entering the religious realm, but here goes. I just ask you to at least respect this view, even if you believe my views to be wrong. After multiple years of research, it's my view that the talk of the world running out of power is pure nonsense. In the U.S. alone, there is enough oil to supply the country's needs for over 200 years. The trouble is that there is no way to legally access many untouched supplies of these natural resources due to environmental moratoriums that only exist in my country and nowhere else. If China can build a new 1000 megawatt power plant per week, then I see no validity to the talk of the world running out of power. For those who live outside of the U.S., just know that there are multple power plants that have been planned for construction that are halted not because of a lack of the resources to run them but because of political reasons. If we can't overcome these hurdles, then enterprise computing will face the perils I mentioned.
3. Automotive market suppliers vs. manufacturers: I believe there is some confusion here, let me further explain. My article's mentioning of the automotive industry references suppliers of parts and products to the automobile manufacturers and compares them to the same for storage. The shift to energy-efficient products over the last 30+ years in the automotive industry has resulted in a shrinkage in the suppliers of parts and products to the auto makers by a factor of five, not a decrease in the number of car manufacturers. For more details on this thinking, please do visit the link to the article. I do believe that the storage industry is moving down the same path since there are dozens of suppliers that could likewise disappear as the industry's revenue potential falls. Both industries are very similar in that fewer than ten companies provide the "big iron" (e.g. GM, Ford, etc in autos, HP, EMC, etc in storage) to customers.
Thanks for your responses, I truly appreciate your opinions. If you have further feedback on these comments, please feel free to post them and I'll do my best to respond. I know there must be disagreements with my view on natural resources, so if that's where we differ, let's just agree in advance to disagree.
Just spend more money
the storage industry just needs to get with the program.
Spend more money on new kit. It will be smaller,faster and cheaper.
It will be less wasteful (that's stockholders money you are wasting)
You will provide a better/cheaper service to your customers and so outcompeting your competitors.
There will be a period of consolidation amongst suppliers as those companies who have invested wisely reap the benefits and gain market share
The cost of energy is pretty level across all storage suppliers and is not a significant differenciating factor.
Unfortunately, given that your article says "We need to build more power stations", we must start from the position of "How would we fuel those stations?"
Secondly, given that you appear to think that all power stations run on oil, we must ask the question - How much oil is there?
Current estimates vary from between 20 years to around 100 years for worldwide stocks at current rates of world consumption, depending on whether you're wildly pessimistic or optimistic.
If your "200 years for the US stocks and US market" was accurate, the US wouldn't be in the least bit worried about current oil prices or the middle east and Russia, because they'd simply drill that oil, raise barriers to export and live happily for 200 years.
Where do you think these deposits are?
Alaska is very intensively drilled already.
Therefore, your proposition is bogus as it's based on incorrect information.
China is building loads of new power stations because it's got coal. Those stations burn powdered coal.
China is not going to export the coal it needs to run those stations, therefore anything it does is irrelevant to the US industry.
While it's true that there are already diminishing returns in absolute terms, as the cost of energy increases (due to the aforementioned decrease in core energy supplies) the actual value of those returns will increase.
There is an energy floor for storage - but we are several orders of magnitude away from it.
Gods! It's really not that hard
- Price the carbon dioxide in the energy.
- Price the energy in the storage.
- Let the market do the rest.
Pollution, in all its forms, occurs because it's cheap to do. It's cheap because the effects are externalised to the economy shielding the producer of the waste.
If you're worried about transition, then ramp it in. Once you've got them by the wallet, their hearts and minds soon follow.
Shortsighted, too narrowly focussed.
Firstly I must admit, I'm not bothered if the data center industry keeps growing or not. It's probably a good idea if it does stop growing at some point. I uses a lot of energy and wether we've 10 or 10000 years supply of oil left I think, if it comes time to make a call it's more important to have the energy to produce food and medicine than to keep the decadent west (and growingly decadent east) in World of Warcraft and Warez servers.
In reality though these things that aren't actually mutually exclusive. You can spend money on both energy efficiency and lobbying for drilling in Alaska if you want, industries that focus exclusively on one or the other are most likely wrong. So try and get more energy produced if you want but I think you're wrong to contend that spending money on efficiency is a bad idea for storage/server makers. Even if it's not a great idea for them it's a good idea for the rest of the world. Maybe lobby for some subsidy of the R&D they're doing instread of advising them not to do it?
I'm no convinced it is a bad idea for them though. Energy being needed, as it is, for industry shurely more industry can happen per KWH with more efficient design & manufacture. R&D isn't cheap, but multi MW powerplants and the oil/gas/uranium it takes to power them certainly aren't fucking cheap are they? And given that the price of energy seems likely to increase massively over the next few decades it makes this R&D even cheaper in comparison.
Plus, there's a multitude of ethical issues that are ignored here.
Even if there is and extra 500 years of oil underground is it not grossly irresponsible of use to piss our way through it as fast as we like? As custodians of the planet does it not fall on us to figure out the absolute best and most efficent ways of using this STILL FINITE resource, or do you think the humans of the future won't want plastic and fuel?
If we suck it all out and burn it as fast as we can we're just going to grow our populations beyond what the soil itself can sustain and have an even bigger problem when we run DO out. And the fact is we ARE running out of this stuff, how fast is seemingly open to debate but clearly we don't have a perpetual supply of it. Massive amounts of energy are going to be needed to build the sustainable generation capacity we are going to need in the future (whether it be in 25 or 250 years) - pissing it away to prop up server sales and the data-center industry may reduce our chances of actually getting this infastructure built in time to avert an enormous problem.
Also, don't forget the desktop/portable computer in all this, there's a synergistic knock on from the server world's focus on 'green' computing, or is making that use less energy in the domestic sphere a bad idea too?
PS: @But you don't see hippies telling us we should stop using Aluminum.
Erm, yes you do, and in some cases we probably should. As noted it's very electicity expensive to refine and is therefore (unless it's lightness leads to great energy savings later) a bad idea environmentally.
I think the internet is great...
It is the best information resource ever invented, and is a brilliant means of communication.
However, 99.999% of the energy it consumes adds 0.001% of its value. All those endlessly downloaded u-tube videos of people being prats - who would seriously miss them?
Regardless of what others say, the world is in the middle of the final energy crisis. Power is the rate of flow of energy. The US may (or may not) have 200 years of oil (at current consumption rates) under the ground, but digging it out (because oil shale is DUG not pumped out of the ground) is VERY expensive and is rate limited by the supply of other resources, not least human. As all forms of fossil energy get harder to extract, we either spend more and more money getting them, or we accept a lower flow of energy - ie. less power.
Why does it take a modern PC longer to do the simple word processing than an 8086 based machine did 25 years ago? It's processor is nominally 100,000 times faster (and more power efficient). We need to reinvent the internet to use the available resources more efficiently. If we got 1% of useful data out of the bits flying around the world would be transformed!
More, more, more...
This might be a bit controversial, but does anyone know where all this extra data is coming from? Really?
Every year we seem to need to double our GHz and GBs in order to do *the same amount* of business. It's high time we pulled all those old coders out of retirement to teach the kids how to work within constraints rather than the seemingly limitless (computing) resources that seem to be available these days.
We will use more power in the future
By 2100 humanity will be using power at much more than the current rate and it'd better be low-neutron fusion.
But storage can be much more energy efficient than currently.
How about dark storage? Your most recently used 10 TB is on disk and everything else is shuffled off to banks of SSD that sit powered down 99.9999% of the time.
This would be especially useful for "write only" storage like web logs, blogs and the UK's constant surveillance of all subjects of the crown.
This dark storage would then only be accessed if say the NHS needed to check how many times you'd used any bathroom in the past ten years.
A third way
There’s no doubt that if you stop innovating, in any sphere of technology or science, you will at some point hit a brick wall and the laws of diminishing returns will kick-in. So Steve Denegri is right, somewhat; if you take a 2 dimensional approach and exhaust the potential of more efficient kit and/or more effective virtualization, at some point you’ll need to start buying more servers, more storage and more juice.
However there is a third way; how about actually managing your data more efficiently and effectively so you don’t need all that horsepower in the first place. Whether it’s realising that the kitten videos in your inbox (as a previous post mentioned) don’t need to be backed up 6 times, or more seriously implementing archiving strategies that recognize 85% of data in production databases is inactive and could be relocated to highly compressed archive stores. I’d contend that new and under utilised data management technologies that allow you to store 30x-40x more historical data on existing infrastructure can have a serious impact on power consumption. A new innovation that should at least be considered before we start building those new datacenters and power stations.
Re: Good Responses
"So most of you seem to be saying that the storage industry should fix it's problems by telling it's customers to use fewer of their products. That'll go over well with the stock holders, I'm sure."
Well, we aren't the shareholders, are we? We're the CUSTOMERS.
Why the feck should we care if the shareholders are unhappy? Harpsichord Manufacturers' shareholders weren't all that impressed when the sales went down either.