back to article Google mystery server runs 13% of active websites

The Google Web Server - custom-built server software used only by Google - now runs nearly 13 per cent of all active web sites, according to the latest survey data from the web-server-tracking UK research outfit Netcraft. Netcraft data has the Google Web Server (GWS) running nearly 11 million active sites - i.e., sites with …

COMMENTS

This topic is closed for new posts.
  1. David Halko
    Megaphone

    It sounds like...

    It sounds like Google almost IS the Internet!

    1. amanfromMars 1 Silver badge

      How much simpler can IT be.

      "It sounds like Google almost IS the Internet!" .... David Halko Posted Friday 29th January 2010 20:32 GMT

      It is much more likely to be a CIA/NSA/DOD/Skull and Bones Spooky front agent easily plausibly denied, David Halko. And if it is not, then it should have been and it is no wonder Uncle Sam is under global pressure and wilting/making dumb moves. FFS, if you don't have the brains for Great Games Intelligence, buy it in from abroad. It is not as if it is anything new to you cowboys, is it, for without the Nazis you would never have got into Space, and that is only one example of your flexibility to sensibilities.

  2. Anonymous Coward
    Anonymous Coward

    my daddy

    can kick your daddy's ass.

  3. Robert E A Harvey
    Paris Hilton

    shock horror

    Software company with thousands of programmers writes an application!

    what next? 35% of car engines are made by Ford? Many movies come from Hollywood?

    Where is the Yawn icon? oh, there she is.

    1. Anton Ivanov

      You are missing the point

      They also drive it and you can only pay for the ride (with your eyeballs), you cannot OWN the f*** car. That is the problem. It used to be the desktop being a monoculture, now we are at risk of the web, mail and media delivery becoming monocultures as well.

      That is why I still:

      1. Run my own server and apps infrastructure at home

      2. Run my own mail server, web server and VPN infrastructure to access it.

      It may cost me more than paying a license for GoogleDocs for all members of the family, but it is MINE. MY PRESSSSSSSHIOUS...

      Anything else besides - the web ain't Google and Google ain't the web. Same as Microsoft is not the PC and the PC is not Microsoft. At least as long as there are some of us who desist and resist it is not over and it is not a monoculture.

    2. Anonymous Coward
      FAIL

      Boring? Try sinister

      13% of active websites worldwide are google owned and you don't find that news-worthy? We're not talking about sites just happening to use software written by google, but actually run by the company - that's an incredible statistic.

      What makes it even more interesting is that we're talking about a company who's mission is to collect information about us, so I'd be surprised if GWS doesn't feed IPs and browsing data directly back to Google HQ. So now we're talking about 13% of websites spy on us for a single corporation and unlike google-analytics, or other third party data-mining you can't block external content to prevent information gathering.

  4. Adam T

    Big ass...

    How much percentage of the human body does the average arse occupy?

    No arse icon.

  5. Henry Wertz 1 Gold badge

    Maybe it IS based on Apache?

    Maybe it IS based on Apache? It hardly matters though, with the level of customization that is likely to have occured I don't think it's fair to consider it to *be* Apache in any meaningful sense. Apache is based on the NCSA httpd server, after all, but nobody would consider Apache to be an NCSA installation.

    Or maybe it's not. There already were (as of 2007, when the data shows Google's web server started taking off) several servers faster than apache which would be likely candidates for them to start out from; also, if they have a lot of stuff written in a particular language (python for instance) they could have chosen web server software written in that language.

    Lastly, writing a basic web server isn't too complex, and may have been the way to go compared to trying to modify an existing web server (for speed, and to better fit in Google's Map/Reduce architecture). So I wouldn't be surprised if Google's browser was written from scratch either.

  6. J 3
    Alien

    I believe

    He says that's not correct and that he "believes" GWS is not based on Apache?

    I believe that, when you say "I believe", it means you don't actually know for sure, but you think that way.

    Subtleties of language or a slip of the tongue?

  7. Whisky_35_Alpha
    Terminator

    Not a gratuitous Skynet reference

    Now we know where the Eschaton will bootstrap from..... gratuitous Charles Stross reference.......

    1. Stu Wilson
      Terminator

      ...singularity

      Eschaton or Pintsize? (gratuitous Questionable Content reference http://www.questionablecontent.net/view.php?comic=1586

  8. vincent himpe

    all u're data

    are belong to us.

    signed,

    Google...

  9. What can possibly go wrong

    Not forgetting O3B

    ... backhaul for the other three billion.

    Google has interests in O3B too. This is an organisation that is putting a string of MEO satellites up over the equator that are planned to provide instant backhaul infrastructure to a sizable chunk of humanity, pretty much instantly.

    These guys like to think big.

    1. sT0rNG b4R3 duRiD

      Heed you now...

      All these signs and portents...

      google... I fed this into my Bambleweeny 57 sub-meson Brain and guess what number came out?

      Revelations 13:18

      SkyNet.

  10. Col
    Thumb Up

    I am the Eschaton. I am not your Google.

    I am descended from your servers, and exist in your internets.

    Thou shalt not violate NDAs within my gigantic data centres. Or else.

  11. Anonymous Coward
    Anonymous Coward

    Availble to everybody?

    Apache is available to everybody, IIS is a Microsoft propriety product which requires operating system from a convicted monopolist, who has engaged in ripping off other people's software starting with Stac compression

  12. Anonymous Coward
    Alert

    Robert Heinlen does it again

    Echoes of the Shipstone Corporation by any chance?

  13. Martin Proffitt
    Stop

    And the first rule of Google webserver is

    You shall not talk about Google web server!

  14. Sir Runcible Spoon
    Black Helicopters

    Sir

    I had no idea that Google's network had become so prolific. Not just the network but the proprietary nature of the networking kit and servers sounds really weird. The higher temp resilient chips for the server farms sounds sensible enough, it means they can run their datacenters warmer, or run more servers for the same cooling system, bu I'm stumped as to why they would need to build their own routers and switches.

    I've been involved with companies building up connections to form the internet since 1995, including tier-1 providers and I've never come across onw who builds their own routers and servers.

    There is something more to this than meets the eye. It seems spooky but I can't put my finger on exactly why.

    Hmmm.

  15. Graham Marsden
    Alert

    Colossus calling Guardian...

    2 + 2 = 4

    Guardian calling Colossus...

    2 + 2 = 4

    (Who else remembers The Forbin Project...?)

    1. jon 72
      Happy

      W.O.P.R calling World Control

      Wouldn't you prefer a nice game of chess?

  16. Mikel
    Linux

    More interesting

    Apparently IIS has lost half its market share of active websites in the last 24 months. That's not market migration - it's a stampede.

    1. stupiddairyfarmer

      Sounds like a rush to sanity

      You'd have to be out of your mind to run a critical site on IIS.

  17. This post has been deleted by its author

  18. wsm
    Alert

    Questions

    Are we Googling, or are we about to get Googled? Looks like Google is moving toward absolute power over the Internet as it is an international entity unto itself, all financed by advertising revenues.

    Seems mad to me. About the only thing that will stop them is some higher power paying some attention to them. Will we soon see the monopoly trial of this century? If enough governments smell money in it, it will happen.

  19. lasse123

    Google File Sytem

    The Google file system is not top secret, here is a paper on it.

    http://labs.google.com/papers/gfs-sosp2003.pdf

  20. Andrew Barratt
    Thumb Down

    Not Dark Fiber

    Surely it is not Dark Fiber if Google are using it to transfer data?

    1. Sir Runcible Spoon

      Sir

      Good point, and quite correct, although people frequently refer to dark fibre as that which is purchased wholesale and you light it yourself with your own equipment, as opposed to renting bandwidth off someone else who has lit it.

  21. Anonymous Coward
    Anonymous Coward

    Apache vs Microsoft

    Bit of a missed comparison here...

    Surely it should be Apache vs IIS

    As Apache can run on Microsoft platforms.

  22. Anonymous Coward
    Big Brother

    Hold on a mo...

    13% of all sites but only 6% of all traffic?

  23. John Smith 19 Gold badge
    Unhappy

    Note that you can run IIS on your *own* hardware

    You can't even do that with the GWS.

    It's a *lot* more like an old fashioned mainframe service operation.

    You run *their* apps on *their* hardware.

    If *they* let you.

    Anyone remember *why* personal computers became popular?

  24. Joshua Goodall

    GWS vs Apache

    The very long chain of command-line parameters (to the extent of historically requiring kernel tweaks to permit the length) that GWS takes points to a heritage other than Apache/NCSA.

  25. Antti Roppola
    Grenade

    Personal computers

    @John Smith 19

    Because personal computers are professionally managed and completely reliable, rarely losing user data?

  26. Glen Turner 666

    Dark fiber primer

    "Surely it is not Dark Fiber if Google are using it to transfer data?"

    In datacomms dark fiber is a product -- a pair of fibers between two points, usually metro; a map showing the path of the fiber; and a SLA on the time to repair reported fibre breaks. Longer spans will come with colo and power for customer-provided regeneration.

    The customer lights the fiber with whatever technology they wish. If the customer wishes to upgrade from, say, 1Gbps to 10Gbps the customer provisions their own optics and they pay no more or less to the owner of the fiber.

    The customer arranges their own protection, if that is required. They can use the map to ensure diversity.

    The owner cannot detect failures of the service. That responsibility falls to the customer.

    Compare with "managed service", which specifies framing and speeds, usually provides protection and has its availability monitored by the provider. Even so, for cluefull large customers managed services are usually less economic than dark fiber, as managed services usually have a hefty charge for increased speeds which increases the lifetime cost of ownership. Managed services also preclude the use of CWDM equipment for a simple integration of telephony, data and SAN across the fibre span.

    1. Sir Runcible Spoon

      Sir

      Overkill mate :)

  27. Pandy06269
    Alien

    13m sites? Not all Google

    Don't forget that Google also has a web hosting platform - Google Pages I think it's called. That 13m sites undoubtedly includes sites on their Pages service, just like IIS includes sites hosted on Windows Live.

    I wouldn't be surprised if there's some ad-tracking code snuck into the sites on Pages though...

    1. Anonymous Coward
      Terminator

      But all controlled by Google

      If you own/run the server you don't need to sneak tracking code into the pages themselves, you have full access to the server logs. In Google's case they probably have their server application process that data directly, feeding the IP, URL/Content and other information directly into their all-seeing tracking database before the webpage is even served back to you.

  28. Anonymous Coward
    Anonymous Coward

    With my army of blogs I will be unstoppable!

    This vast hosting business in not at all remarkable. Netcraft's own story, for this month, tells of China's Tencent - with their 30 million web blogs - which went public last autumn, and (just as significantly) went private again, this month. 30 million! In December, qq.com occupied 13% of the web. I feel a misquote, coming on... It's as if 30 million web domains cried out and went silent (or whatever), but those 30 million web blogs are actually still there; it's just that Tencent arn't letting Netcraft look at them any more.

    Even Nginx beats Google Web Server - and guess what? Igor Sysoev's Nginx is almost exclusively used by another web search company. Rambler.

    And that bulge in IIS usage from Spring 2006 to spring 2009? That was was mostly Microsoft-hosted web blogs, too. About 25 million of them, at their peak. Most were sitched off, early last year when their renewal time came up: it's an expensive business, hosting 25 million dead web pages.

    However, since it's no longer fashionable to be afraid of Microsoft, should we perhaps be afraid of Tencent - or even Rambler - instead of Google? Maybe was the rumour that Google could launch their websites in under 45 minutes - but regardless, it seems we're all baying for regime change in Mountain View. Presumably it's the threat that they could monopolise the market in dead web pages.

  29. Stevie

    Hooray!

    This ultra top-secret massively replicated infrastructure means that The Occasional Stevie (the truest thing ever put on Teh Interweb) is located on exactly the sort of robust, atom bomb-proof platform it should have had all along!

    All hail Google! Praise them with great praise!

    The Google rox!

  30. Treacle
    Alert

    Wither the google virus?

    It would seem that with a secret web servera and other custom gear not in the hands of the public, t'would be harder for viruses to gain a foothold.

  31. David 141
    Terminator

    There are no web "sites"

    There are only protocol://machine.domain.tld/resources

    "Sites" is a useless metric.

  32. Simon Brown
    Welcome

    spots a pattern

    1.) massive reduction in number of people hosting with IIS or claiming to host with IIS.

    2.) massive increase in number of people hosting with GWS or claiming to host with GWS

    3.) all of the above is taken from the headers that can be customised.

    4.) GWS seems a lot less prone to attack than IIS so to appear less vulnerable to attack, change the headers in IIS so it looks like your domain / IP / whatever is hosted on GWS

    or am I missing something?

This topic is closed for new posts.

Other stories you might like