Google, Microsoft Escalate Data Center Battle

miller60 writes "The race by Microsoft and Google to build next-generation data centers is intensifying. On Thursday Microsoft announced a $550 million San Antonio project, only to have Google confirm plans for a $600 million site in North Carolina. It appears Google may just be getting started, as it is apparently planning two more enormous data centers in South Carolina, which may cost another $950 million. These 'Death Star' data centers are emerging as a key assets in the competitive struggle between Microsoft and Google, which have both scaled up their spending (as previously discussed on Slashdot). Some pundits, like PBS' Robert X. Cringley, say the scope and cost of these projects reflect the immense scale of Google's ambitions."

Comments

Hmm... It's Slashdot so...

[gQuigs]

Microsoft's is to run Vista. While Google's is to save the world.

Re:Hmm... It's Slashdot so...

[DJCacophony]

I knew Vista's hardware requirements were high, but a $550,000,000 data center?

Re:Hmm... It's Slashdot so...

[Stephen Samuel]

Don't worry. You don't need to spend that much if you turn down the quality settings.

Re:Hmm... It's Slashdot so...

[dangitman]

This is Windows, so the only quality setting available would be zero.

Non-local computing

[bigberk]

The aim for both of these giants is to shift people towards non-local computing, that is software and applications that run remotely rather than on someone's own computer.

Early signs of this beyond the obvious google applications that require web access, are aggressive attempts by Microsoft to "activate" everything online. You are going to increasingly need network connections to run standard applications.

I don't like that myself, since it hurts reliability and autonomy in computing. From a marketing perspective, there are huge benefits to centralized computing of course. Take gmail for instance, which lets google mine your private communications to gain insight into products and services which might interest you.

Re:Non-local computing

[Speed Pour]

The aim for both of these giants is to shift people towards non-local computing, I thought the aim was to prove which one had the larger penis?

I don't like that myself, since it hurts reliability and autonomy in computing. From a marketing perspective, there are huge benefits to centralized computing of course. Take gmail for instance, which lets google mine your private communications to gain insight into products and services which might interest you. On a serious note. While I don't care all that much if google uses an automated method to push advertising on me, I am more bothered by the fact that it's a single target that retains tons of information. A hacker can break into one person's home computer and get their info, or they can break into a google server and have 2 million people. Same reason that hackers target windows/ie over linux/firefox, they can accomplish/demolish a larger audience.

Re:Non-local computing

[solitu]

Microsoft desperately needs new datacenters because their search index size is in need of an increased capacity. Google with its 100000++ computers is able to record every single click-through, record your chats, store your email for posterity (even after you delete it), store every single search query for several years, record your online transactions etc. not only on its own sites but other sites like slashdot for example. This has helped improve their search result and provide targetted ads among other things. Microsoft's search now algorithmcally matches Google. It now does a great job for most queries, but for some esotoric queries its small index size is very apparent.

Re:Non-local computing

[mcrbids]

Early signs of this beyond the obvious google applications that require web access, are aggressive attempts by Microsoft to "activate" everything online. You are going to increasingly need network connections to run standard applications.

I don't like that myself, since it hurts reliability and autonomy in computing.

If all else is equal, a centralized approach is less reliable than a distributed approach.

But seldom is all else equal.

A distributed approach to software and information systems often has catastrophic failure as part of the mix. A well-designed central approach, with built-in redundancy and a qualified backup scheme can usually outperform the poorly administered "edge" systems run by end users.

And, in this space, the economies of scale rapidly factor in, making a better experience cheaper, as well. Sorry you don't trust the hosting providers, but it isn't always that way...

Re:Non-local computing

[zCyl]

I don't like that myself, since it hurts reliability and autonomy in computing.

Don't worry, you can trust skynet. What could go wrong?

Re:Non-local computing

[Richard W.M. Jones]

A hacker can break into one person's home computer and get their info, or they can break into a google server and have 2 million people.

I'd be more worried about a rogue government or future government deciding
they want to mine that data to find out who all the "terrorists" are.

Oh, wait ...

Rich.

Re:Non-local computing

[totally bogus dude]

A well-designed central approach, with built-in redundancy and a qualified backup scheme can usually outperform the poorly administered "edge" systems run by end users.

True, but a) you have no idea of knowing just how resilient their systems are, or how reliable their backup scheme is... until it fails, of course; b) online apps require an internet connection; and c) trust.

The need for an internet link to the central site is still a pretty significant failure point, especially if we're talking "end user" systems which are probably connected via a single phone or cable line.

Trust is probably the most significant problem. Not just that the company that stores your files will do so in a secure and discreet manner, but also that they'll behave in an ethical way. Once you become reliant on a service, they can start extorting you for access to your own documents. They can increase their fees, and refuse to release the documents to you until you pay them a severance fee. They can then release them in a secret proprietary format which only their systems can accurately interpret. All of these things you could sue for... but do you really want to be suing a monster corporation (or even a small, nasty one) to get your own documents back?

And what happens when your favourite do-no-evil corp is bought out or sells their central application services to a do-nothing-but-evil megacorp? Quickly grab a copy of all your stuff and then delete it? How do you know it's actually been deleted?

</doomsday>

Good in the short term

[caitsith01]

For the time being, it's surely a good thing if two extremely wealthy companies pour resources into creating ultra-high capacity facilities such as these, particularly as Google's business model is based around providing services which are nominally 'free' (in terms of dollars) and as such these resources are in a sense an investment in our common infrastructure. If we're really lucky Google and Microsoft will hugely over-invest, and one day find themselves with a large overcapacity which third parties might be able to use for their own work.

However, longer term things may not be so appealing. Both companies have a nasty habit of collecting and storing as much personal data as possible (Google in particular), and both are pushing towards 'lock out' where you are prevented from using your own computer without their participation via connection to their networks. And of course the software industry has a history of producing only one winner in the end, meaning the benefits of this kind of head-to-head competition are unlikely to last...

Re:Good in the short term

[rumith]

From my point of view, there'll be no single winner, but technology will revert once again, and the term 'computer' will mean what it meant in the 60s and the 70s. Provided enough bandwidth, stability and solutions like roof-top server rooms - Google [or Microsoft, although it's hard for me to believe it] has good chances to build such a network with powerful data centers and relatively dumb clients. Again, the task is not easy, and there is 1001 reasons why, but defying laws of physics isn't among them, and the Almighty Buck will surely help solve all of them sooner or later.

If we're really lucky Google and Microsoft will hugely over-invest

Why? Google's desperately trying to diversify its income sources, why don't you suppose that they'll offer hosting services because they plan to?

and i quote

[macadamia_harold]

These 'Death Star' data centers are emerging as a key assets in the competitive struggle between Microsoft and Google

That's no zune...

Time to invest

[Technician]

On Thursday Microsoft announced a $550 million San Antonio project, only to have Google confirm plans for a $600 million site in North Carolina.

It looks like it's time to invest in IBM, Red Hat, Maxtor, and Intel. They may sell a lot of hardware and software.

Data security nightmare

[Travoltus]

If it isn't the hackers trying to break into your system, it's Google's marketing partners getting exclusive access to your communications.

Forget that, I'd rather have my own mail server at home, not to mention my own apps at home. I don't even trust ISP's.

This "offsite word processing" crap is for chumps - anyone with sensitive data would be utter idiots to go there.

Ecological nightmare

[Duncan3]

So now we know why the sky is always black with pollution in sci-fi movies... we cover the earth with multi-gigawatt eating data centers.

Since electricity is a continent-wide commodity you can guess whose electric bill will be going up as they buy up all the watts just so they can store every little detail about your life.

Death Star independent contractors

[MillionthMonkey]

A construction of a real Death Star data center would require a lot more manpower than Google or Microsoft has to offer. I bet there are independent contractors working all over these things: plumbers, carpenters, electricians, DBAs, MBAs, roofers, etc. In order to get one built quickly and quietly they'd have to hire anybody who could do the job. Do you think the average Google employee knows how to install a toilet main? All he knows is JavaScript and Knuth.

All these independent contractors in each Death Star data center are getting involved in a war between Microsoft and Google- a war they had nothing to do with.

No they both need those data centers

[Anonymous Coward]

Google needs a $600 million data center to serve its 300 million daily users, while Microsoft needs its $550 million data center to serve both of it's MSN Live Search users.

ObStarWars

[OldManAndTheC]

These 'Death Star' data centers are emerging as a key assets...

Better make sure to protect the plans for that data center...one well placed shot in an exhaust vent could take out the whole thing. Not much harder then hitting a womp rat with a T-16, from what I hear...

Speaking as an independent contractor myself

[Anonymous Coward]

I'm an engineer, and I can tell you that an engineer's politics come heavily into play when choosing jobs. Just three months ago I was offered a job building a huge data center in the valley, in a vast facility. And then I learned how screwed up the company's financials were. The money was right, but the risk was too big. So I passed the job onto a friend of mine.

They just laid his ass off and shut down the entire outfit, but they still have to run the air conditioning because of a few third party servers left over. He wasn't even finished running his CREATE TABLE scripts. I'm still employed because I recognized the risks involved in working in a Death Star. Anyone working in a Death Star data center for Google or Microsoft is aware of the risks involved in that war. Whatever happens to them is their own fault.

Re:Speaking as an independent contractor myself

[JoeCommodore]

This needs a rewrite:

I'm an engineer, and I can tell you that an engineer's politics come heavily into play when choosing jobs. Just three months ago I was offered a job working at one of the big information companies data centers, in a vast facility. And then I learned how screwed up the company's plans were. The money was right, but the risk was too big. So I passed the job onto a friend of mine.

While writing a C# script for some part of thier web portal my friend was hit by a flying chair, it was a leathal blow, and he died instantly. I'm still employed because I recognized the risks involved in working in a Death Star. Anyone working in a Death Star data center for Google or Microsoft is aware of the risks involved in that war. Whatever happens to them is their own fault.

As long as it doesn't violate GPL

[melted]

As long as it doesn't violate GPL (and it does not), I'm fine with Google not releasing their stuff to the masses. Nearly every big Linux shop has their own tweaked version of Linux kernel, so it's not like they're evil or something.

Re:As long as it doesn't violate GPL

[John Nowak]

Would you prefer Google not exist at all or be forced to strike some deal with Microsoft?

Re:As long as it doesn't violate GPL

[chris_mahan]

If you avoid code sharing and community reviews for long, you end up with a sub-par, brittle, expensive and proprietary solution that costs more than it earns. You ignore the great unwashed hackish masses at your own grave peril, O Googole.

Re:As long as it doesn't violate GPL

[dangitman]

It just seems to defeat the open source nature of Linux when you branch in a private way that avoids community code review and source code sharing.

If it's against the spirit, then why was private code-branching specifically allowed by the GPL? Isn't freedom to run your code as you see fit a big part of freedom?

Re:As long as it doesn't violate GPL

[bmo]

"It just seems to defeat the open source nature of Linux when you branch in a private way that avoids community code review and source code sharing."

It's obvious that you've not grokked GPL itself.

The GPL covers distribution. No distribution = do whatever you want with the code.

You forget that Google loses the power of peer review for their code, but that's the tradeoff. Having a lot of really smart people in their employ probably makes up for it. So they've got their own branch. They have to do their own heavy lifting.

If you remove the freedom to work on Linux in-house, then you've removed one of the freedoms _allowed_ by the GPL.

--
BMO

Re:As long as it doesn't violate GPL

[tacocat]

I don't think it's a violation just as long as they keep it in house. Which means they also have to support it in house. Not everyone is willing to keep on retainer kernel developers for their employee desktop computers.

Google is changing the way people do business on the internet. They are also going to change the shape of the internet. Much of this very likely will follow any of a number of historical industrial patterns which eventually lead to severe regulations and a severe restriction of who is allowed to post information on the internet and what kind of information you are allowed to receive on the internet. It is not necessarily true that the regulators will dictate the limits of content but simply reinforce the idea of limiting content.

Examine the history of Television and Radio to see how they followed this path. I don't think anyone really considers the internet that much different. At least they can get it to fit the model. With the exception of the social webs like facebook, youtube, and myspace, most of the internet consists of content delivery and a large portion of that content (by some) is seen not as written words but media in forms of video/audio material. And with the highly publicized problems that these social networks are having (where everyone is a pedophile or worse) it's ripe for all the sheeple to cry out that they need the guberment to protect them from their neighbors. And "bang!". Just like that you have a completely "owned" environment where no one can actually do anything, everything costs money, and the sheeple are happy again.

Re:As long as it doesn't violate GPL

[hey!]

Tailoring software to your own use is not branching. It's just using. A "private branch" is a contradiction in terms. Perhaps it could be called a "private bud", because such a modified version could become a branch. But if it can not attract users and developers, it's not a branch.

The question of the "spirit" of FOSS is profound though. Underneath the license, there are two related principles, a negative one (do not interfere with the rights of an object code recipient) and a positive one (share knowledge). The question that arises is this: should these principles apply to users of services built around the object code? There doesn't seem to be a fundamental reason why such rights are granted to people who receive the object as object code, but not people who are equally if not more affected.

I think the answer may hinge on this: of the two principles, non-interference and sharing, the sharing principle is less strong.

Users of a service created by a vendor like Google are not supposed to have the power to change that service. Otherwise it would be impossible to offer a service before its users redefined it into the oblivion of inconsistency. Google gets to define the service and control it. Not allowing users to change the service (via the source code it runs on) is not interference, because the service would not exist if any user could change the source code on a whim (Wikipedia perhaps being a related counterexample).

But if the sharing principle were equally strong Google would be obligated to share the source code of any changes it made with its users, even if they were not allowed to alter the services they depend on.

This argument leads to the conclusion that sharing must be less of a fundamental value to FOSS than it is "instrumental" to the value of non-inteference. If you control source code to object code somebody else depends on, you can interfere in their freedoms (e.g. proprietary database licenses that forbid publishing benchmarks).

This may make some sense. In engineering, the most important piece of knoweldge is usually that something can be done. In this case, the changes Google has made are probably (1) stripping unneeded features out and (2) tweaks that are highly Google specific. The first is something that any reasonably competent engineer can do, the second is probably not critical to any would be competitors amongst Google's users.

Control over source code is reaching, via the laws of copyrights and contracts, into the affairs of object code recipients. Non-sharing of know-how is something every business does to some degree; it is more difficult to draw the line between vicious and innocuous secretiveness than it is between vicious and innocuous interference.

Re:As long as it doesn't violate GPL

[somersault]

Funny how lots of other software companies manage it. Google has a lot of smart people, who get paid to code, it's not like it's just some guys who do it for fun in their spare time. Presumably they don't mess with the kernel or anything, and can just upgrade the same as everyone else, maybe with minor tweaks to their own code. And I seriously doubt Google is making less money than it spends, considering it's building $1,500,000,000 worth of data centres. Would you really like it if spammers could view Google's source to learn more about PageRank and scwew awound with your search results?

Death Stars

[clickclickdrone]

FX: Guard on gate waves hand mysteriously 'This isn't the Data Complex you're looking for'

Incentives to Build

[necro81]

Is anyone else a bit weirded out by the massive incentives the local governments have offered. I know this is nothing new, and the locals hope that these will spur further high-tech development in the area, but let's examine these cases:

San Antonio (Microsoft): No property taxes for 10 years. A $5.2 mil grant from the CPS Energy economic development fund to pay for the electrical infrastructure to build the site.

South Carolina (Google): No property taxes for 30 years (essentially, for the life of the site). The 150-acre site was granted to them, and the state government has granted about $5 mil, too. Google has been incentivized to the tune of about $100 million.

Some of the structural construction will undoubtedly be done by locals. The technical work of building the data center (installing servers, wiring everything together) is probably outside of a local construction company's expertise. The real bulk of all those hundreds of millions of dollars goes to purchasing the actual computer equipment, none of which is local. A handful of the most-well-educated locals could be employees, but most employees will be transplanted. In less than 10 years, both sites will probably be obsolete (or, worse, axed as excess capacity). As the article on Google's site notes, the obscene incentives equate to "a $500,000 sweetener for each of the 200 jobs Google will create."

For half a million dollars, I'm sure the local economy could get more bang for its buck than just one Google employee. What exactly are these local governments getting in return for their obsequiousness and prostration?