If you had to list the top manufacturers of computer servers, you would think it would be fairly easy…
- Dell
- HP
- IBM
Google? Yes, that is surprising but true. And even more surprising when you learn that the fourth largest manufacturer of computer servers doesn’t even sell them… they are for internal use only.
This grew out of Google’s humble beginnings, when to save money the founders built their first 
servers from cheap and often second-hand parts. With $9 billion in cash the company no longer needs to pinch pennies, but they have found that home-built computers let them optimize the systems for speed, flexibility… and energy savings. The energy requirements of huge data centers, such as those operated by Google, are enormous. So enormous that they have a group that specifically scouts out real estate near cheaper hydroelectric sources of energy that happen to be near major internet fiber optic backbones.
Martin Reynolds of the Gartner Group estimates that this focus on efficiency has led to Google’s computing costs being half that of it’s rivals and a tenth that of corporate technology groups. Not chump change when a company like Google spends a couple billion a year on computing infrastructure. Alan Eustace, Google’s VP of systems engineering, claims that they can deploy systems faster, cheaper, and at scale larger than their competitors at Yahoo and Microsoft, and that this can give them up to a five year lead in the market.
Of course there are the critics of this strategy, notably Bill Gates himself. Most claim that it’s not a necessary internal competency and that it de-focuses the company from their true mission.
But as lean manufacturers have known for a long time, it’s not all about such a narrow definition of cost. By investing heavily in server manufacturing and design, they have been able to create (or stumble upon) technologies and applications that have given them a competitive advantage. They can then deploy those changes faster, and react to positive and negative feedback faster… similar to how in-house lean manufacturers can react quickly without having to deal with supposedly cheaper parts sitting in containers on ships coming from China.
In the case of Google this has led to the use of lean manufacturing techniques. Urs Hölzle, Google’s VP of operations, is working on ways to optimize reliability and performance. Maintenance is optimized by using velcro to hold server components together… a characteristic of the original home-built servers. As the limits of hardware reliability are reached, software is optimized to take over and automatically spread data risk across multiple redundant machines.
Google is even looking at creating its own high performance silicon and chip designs. Definitely not what we’d think a typical internet search company is working on. But this internet company knows more about manufacturing than many manufacturing companies.
John Hunter says
“Most claim that it’s not a necessary internal competency and that it de-focuses the company from their true mission.”
Defocuses – possibly. I don’t think so, but at least it can be argued.
“Not necessarily a competency.” Who does it better then? Who provides the computing power Google needs better than they do? Nobody. Definately it is a competitive advantage for them. It might not have been obvious 2 years ago that investing in this part of the business made sense to most of us (though to the founders I am sure it was) . But it seems pretty obvious to me now.
Karl McCracken says
It’s also a very kaizen-like method of ensuring competitive advantage.
If you go and buy a big swanky system from someone else, then provided that your competitors have deep pockets, they can easily copy you. This happened to a firm I worked for a few years back – Viasystems in South Tyneside, UK. They were completely blindsided by competitors who basically bought the same equipment and made the same product, but at a much lower labour cost.
With Google’s approach, their data centre performance is developed by hundreds of small in-house improvements. This is almost impossible to copy – even if you try poaching staff on a wholesale basis, the path dependancy of the sum of the improvements makes them highly resilliant to imitation.
So well done Google – a money saving idea turned into a competitive sledgehammer!
Tim says
Kevin,
Great article. I found several others referenced to this same topic as well.
* You might want to get your side-kick, Mr. Bill to read about this discovery.
It once again proves him wrong and myself to be correct, regarding his whole
” Economy of Scale deathbed ” rant.
Google is DEFINITELY utilizing EofS in many ways. Let’s hope they keep it up.
Thanks again Kevin !
Graham says
Tim: “Google is DEFINITELY utilizing EofS in many ways”
I disagree – they specifically stated the fact they enjoyed that they could treat each server as unique, not as some bland mass.
They would carry out the activity when it was for one pc, they’ve just carried on the trend.
Barry "aka the Hillbilly" says
Looks like no matter what it is Tim sees Economy of Scale in it. Whether it’s Airbus just scaling up an Airplane or Google.
Google has taken one out of the Toyota production system playbook and made equipment that actually fit their needs. Toyota abondoned buying what the Industrial Equipment people were peddling and created a strong in house production engineering group to build or modify their own equipment. Making things the size appropriate for single piece flow. Shingo describes this in some detail in his book on Non-Stock production.
Tim says
Sorry boys, you’re incorrect on this one as well. You might want to take some time and actually read the base of information that Kevin gave us.
It’s in the second paragraph, i.e. the ” surprising but true ” link.
If you had read the article,
titled ” A Search Engine that’s Becoming an Inventor ”
you would have seen the references to Scale Economies, achieved by Google,
as pointed out by Hansell and Markoff, the authors.
OR, maybe you need further proof. How bout Steve Pearlstein’s article in the Washington Post from last November, which was titled, sorry guys, but here it is in plain English, ” Economy of Scale Might Inspire Companies to Ditch IT Departments “.
You can actually use our friends at Google to find this jewel. If that is not enough, maybe you’ll make time to research further, part of his work, where he gives credit to Nick Carr’s article from many moons ago, in the May 2003 issue of the Harvard Business Review.
In his segment titled ” IT Doesn’t Matter ” Mr. Nick goes even further into the subject, describing the Economy of Scale operations currently being enjoyed by many IT departments worldwide. For a night-cap, you can then read another masterpiece by Michael Porter, of the Harvard Business School.
In Mike’s work called ” Five Forces ” he TOO correctly describes the Economy of Scale techniques, as they relate to gaining competitive advantages, in the world of IT.
Therefore, always remember to do your homework BEFORE you post, not after.
Have a nice weekend.
Graham says
>Sorry boys, … It’s in the second paragraph, i.e. the ” surprising but true ” link.
>If you had read the article … you would have seen the references to Scale Economies, achieved by Google,
****************
“Larry Page and Sergey Brin, built their own computers out of cheap parts meant for personal computers. They wanted to save money”
-No economy of scale there.
“most of the hundreds of thousands of servers it will deploy are being custom-made based on Google’s own eccentric designs.”
-No economy of scale there.
“They are building an enormous computing resource on a scale that is almost unimaginable.”
-No economy of scale there, just that they are a very big company.
“We spend a little bit more per machine.”
-Not a very good economy in that scale, eh?
I believe that you confuse the fact that they are a very large company with the idea that they do everything via economy of scale. Can you correct me?
****************
The other articles i’ve not read so i’m not going to make any comment on them, nor am i going to comment on the general topic of economy of scale being false. I just thought i’d point out the irrelevance to the above story.
Hope it’s sunny tommorrow
Graham says
I took a moment to read
******************
Steve Pearlstein’s – Washington Post last November – “Economy of Scale Might Inspire Companies to Ditch IT Departments”.
ok, so Steve states that Carr states that there will be economy of scale if everyone centralises their pc usage – great. No evidence.
You might like to consider the starting line about power plants being centralised? They are trying to turn that on its head right now – there are HUGE inefficiences in distributing power across the nation. Minature power stations in every house are being considered – work to improve the efficiency there and you save mammothly – as simple as no longer digging up roads for power lines.
It’s greenpeace and biased but it’s what i found in 2 seconds to add support.
http://www.greenpeace.org.uk/climate/solution/revolution.cfm
“Sun is even setting up an online auction of excess computing capacity.”
-They’ve got overcapacity, it’s because of inefficient operating that they want other people to buy into centralisation.
******************
Actually being able to read Nick Carr’s article: ” IT Doesn’t Matter “, is a bit of a challenge – i can only find commentary on it.
The summary seems to be that he thinks businesses should run on the same platforms of software – localisation and the right tools being designed for the job are a thing of the past. Exactly the opposite of what the likes of Toyota preach.
******************
Michael Porter, of the Harvard Business School. “Five Forces”
“High fixed costs result in an economy of scale effect” – basically build them high and sell them low
Great – shame it’s been proven time and time again that batch production really isn’t the best thing.
He says it himself:
“Since the firm must sell this large quantity of product, high levels of production lead to a fight for market share and results in increased rivalry.”
“For example, in long distance communications roughly 10% of the market is necessary for MES. If sales for a long distance operator fail to reach 10% of the market, the firm is not competitive.”
You might want to try and circumvent that with a bit of ingenuity instead of building more.
******************
Basically it seems to boil down to that businesses want to sell huge quantities of homogenised products to people who don’t care – whereas those who do care keep their systems under control/responsive/ and close to them.
P.S. – IT most specifically is a competitive advantage at my workplace.
Tim says
Graham,
You’ve regurgitated so much that I must now be careful where I step. You asked in your first responsive-rant, if I could correct you ? Sure. No problem. Your idiot’s logic is self-evident.
1.) You claimed I used the terms ” they do everything via EofS.” You would be incorrect. I originally stated how they reference EofS techniques, within their operation. ( No reference to Lean )
2.) In your second rant, you say that Carr’s work describes EofS, and then you state ” great- No evidence.” Again, you would be Incorrect. The research itself references Google. That would be the basis for Kevin’s original article above. Sorry, shit-for-brains. You too suffer from a lack of reading comphrension. ( No reference to Lean )
3.) You also state ” spending a little more per machine ” is not very good EofS. Sorry once again Graham. If you would take the time to revisit Kevin’s article above, via the 4th paragraph, he mentions how they are using the Scaled Economy system to save money. They are NOT SELLING the machines, as their end product. They are UTILIZING them to achieve an efficiency in cutting their operating costs, to 1/2 of their direct rivals and a TENTH of some other corporation’s IT departments. THAT’S the SAVINGS. THAT’S the SCALED ECONOMY being ACHIEVED. Once again you are proven incorrect. ( No reference to Lean )
4.) Then you make reference to overcapacity at Sun. We were discussing the efficiencies of Google, remember ? Not the inefficiencies at Sun Microsystems. Try to stay on topic. ( No reference to Lean )
5.) You also make a lame attempt at saying Porter’s work is somehow shown to be false, since ” time and again batch production really isn’t the best thing, blah blah blah.”
Porter specifically points out, regarding IT,which is the basis for this discussion remember, NOT the Toyota factory floor, Economy of Scale CAN BE a distinct competitive ADVANTAGE. Sorry Graham, you are therefore yet again, proven to be INCORRECT in your observations. ( No reference to Lean )
Here’s a quick bit of advice. This weekend, try to get out in that sunshine you so long for. It might help shine some truth and reality into your shallow fantasy world.
Graham says
Woof, i’ll ignore your insults and ignore your post. We can do without a flamewar here – suffice to say i do not agree with the findings of your points and it’s obviously i’m not going to find a rational discussion.
Tim says
You’re not going to find a rational discussion ? That’s your problem, not mine. Your avoiding of the facts has a direct impact on your personal bias.
Barry "aka the Hillbilly" says
Graham,
Don’t worry about Tim and his Economy of Scale Stuff. That seems to be all Tim knows. He sees Economy of Scale in Everthing.
I personally believe he has some kind of personality disorder. He seems to like writing WRONG a lot. He seems to read whatever he wants to in everything. He accuses others of Comprehension problems when ole Tim seems to see things only one way, his way. He’s one of those one Trick ponies Bill spoke of.
Tim chastised me for reading Ono and Shingo’s books. Tim will never accomplish anything close to either one of those guys. Particularly since this late in life he is just an Industrial Equipment Executive. For Tim its OK to make fun of Onosan and Shingos work while at the same time quoting from the Harvard Business Review. I wouldn’t pay much attention.
Warren Buffett successfully avoided the Hyped up Tech Implosion of the last decade by concentrating upon good businesses. Hype appeals to Wall Street and Marketing people. I suppose because there isn’t any substance to either.
Tim is about the only person in the world who could do such a thing. I doubt seriously that he had much math past alegebra when he was pursuing his community college marketing degree. I bet he doesn’t even know what Sigma stands for, let alone how to go about calculating it.
Taiichi Ono pretty well laid waste to all of Tim’s claim’s about Economy of Scale in his first book. Shingo goes even further in his analysis.
The economy of scale systems that Tim loves so much do not address many of the wastes identified by Ono and they certainly are not capable of producing product of the Quality level of the Toyota production system.
Tim even thinks the new Airbus is just a simple scale up. Should be able to make a lot more profit if we make that bird twice as large. Its just a simple linear equation, just X and Y. My guess is Airbus is going to have a hard time manufacturing that plane. Just in the last week Tim’s Airbus executives started stepping down due to manufacturing delays of up to a year. In the Meantime Boeing is heading in a different direction and building the Dreamliner. Time will tell who wins.
I think it is becoming increasingly obvious every day that the future belongs to production systems that use less resources and are able to respond to customer’s needs quickly. That system is the Toyota production system as employed by Toyota and also Honda. Not Tim’s Economy of Scale systems that are bloated with wastes and slow to respond.
The most ridiculous thing Tim said was concerning Warren Buffet’s performance. Warren has went many years without having any short term cash concerns. Quite the opposite of what Tim implied concerning a short term balance with the long term. Warren is about as long term as you can get. Warren in the last several years has been complaining about sitting on too much Cash and not being able to find undervalued and desirable companies to purchase.
Another note Tim, actually many of the Toyota Plants do have a union. Just not the Toyota plants here in the US.
And even though Tim claimed that only Bill, myself and about 20 other people on the planet were so dedicated to LEAN. I have a lot of confidence that there are also many thousands of Toyota and Honda Engineers who are perhaps much more committed to Lean than either Bill or myself.
Lastly, in this month’s Quality Progress the editor asked if LEAN was just another FAD. Well in my opinion it could well be in America. It would take America to turn it into a FAD. Our companies are so full of Short term thinkers who are MBA’s and Marketers or Financiers that I think it will be very hard to implement. It took Toyota 30 years of Hard dedicated work. Our companies are ran by people who are incapable of that type of committment.
Toyota may have suspected as much and that’s why they let Onosan and Shingo write about the Toyota Production System.
Toyota and Honda have basically been ran by engineering and production management. They haven’t let a bunch of Professional Short term thinkers lead their companies. Honda in particular has a very strong engineering focus.
Looking back over the last 100 years, most of the the value has been created by the Scientists, Engineers and the working class who worked in the factories. There isn’t much that can be attributed to the Short Term Thinkers.
I for one am happy to be on the side of the Toyota and Honda engineers and upper management.
Don’t be fooled by the one Trick Pony Tim. The future belongs to companies like Toyota and Honda. They are also very environmentally focused and many of their Plants are now reducing, reusing and recycling such that they no longer send waste to a landfill.