Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Businesses Data Storage The Internet

As Cloud Growth Booms, Server Farms Get Super-Sized 57

1sockchuck writes: Internet titans are concentrating massive amounts of computing power in regional cloud campuses housing multiple data centers. These huge data hubs, often in rural communities, enable companies to rapidly add server capacity and electric power amid rapid growth of cloud hosting and social sharing. As this growth continues, we'll see more of these cloud campuses, and they'll be bigger than the ones we see today. Some examples from this month: Google filed plans for a mammoth 800,000 square foot data center near Atlanta, Equinix announced 1 million square feet of new data centers on its campus in Silicon Valley, and Facebook began work on a $1 billion server farm in Texas that will span 750,000 square feet.
This discussion has been archived. No new comments can be posted.

As Cloud Growth Booms, Server Farms Get Super-Sized

Comments Filter:
  • Texas? Georgia? Not exactly ideal for cooling costs. What happened to the green data center movement; put it in North Dakota and all the server cooling can be handled by a vent coming in from the outside.

    • Ah, the south... Plenty of sun, wind, infrastructure, and tax breaks.

      "green" is more of a checkbox than an honest mindset.


      isn't that the human/hubris way? As long as you are going after free energy, what does it matter if you use X KWH vs X+1 KWH?
    • by mlts ( 1038732 )

      Only problem with NoDak is that if it gets snowed in, some type of outage can take days to weeks to get fixed if the weather is bad, and with temperatures well into the negative 40s, it takes specialized equipment to fix things. Texas and Georgia have their problems, but generally, the worst one encounters are ice storms which make an area impassible for a few days, and those are relatively infrequent. The occasional snow is manageable.

      Well, the absolute worst are twisters... but that's what insurance is

      • Only problem with NoDak is that if it gets snowed in, some type of outage can take days to weeks to get fixed if the weather is bad, and with temperatures well into the negative 40s, it takes specialized equipment to fix things. Texas and Georgia have their problems, but generally, the worst one encounters are ice storms which make an area impassible for a few days, and those are relatively infrequent. The occasional snow is manageable.

        Well, the absolute worst are twisters... but that's what insurance is for.

        Think again. Georgia and Texas are both hurricane targets.

        A company I used to work in Florida had a hurricane contingency plan that involved shipping a trailer full of mainframe tape files up to Charlotte.

        The hurricane that was supposed to hit Florida swept around an nailed Georgia/South Carolina instead. The truck with the tapes was stranded between 2 flooded-out sections of road.

        Atlanta is far enough inland that few hurricanes would have enough direct force to faze it (excepting something like what Hugo d

        • Think again. Georgia and Texas are both hurricane targets.

          Atlanta GA is not on the coast. It's not NEAR the coast. Hurricanes are not an issue near there.

          Fort Worth TX is not on the coast. It's not NEAR the coast. Hurricanes are not an issue near there.

          • Take a look at what Hurricane Hugo did to North Carolina and where it did it.

            If you come inland from Savanna, Atlanta is no further.

            Besides, a hurricane doesn't need to still be an actual hurricane to really mess things up. One storm came West of Florida, went Northeast through Georgia, came out into the Atlantic, turned back into a hurricane and plowed down into Florida again.

            And THAT was fairly recently.

      • Only problem with NoDak is that if it gets snowed in ...

        By far the biggest problem with ND is the job market. The unemployment rate is 3%, the lowest in the nation. It is even lower for construction and tech workers. Wages are sky high. Good luck getting your data center built and staffed.

    • by LWATCDR ( 28044 )

      Depends where in Georgia and Texas and they could be using Wind for renewable power in those locations.
      Frankly the place that I would think is logical is in Tennessee using all those TVA dams for power and the lakes for cooling.
      What I do not get is why the heck is anyone building one in "silicon valley". The costs of power, land, and cooling would seem to be very high.

      • by cdrudge ( 68377 )

        What I do not get is why the heck is anyone building one in "silicon valley". The costs of power, land, and cooling would seem to be very high.

        Equinix is a colo. They don't care what the energy costs are as it's just passed on to the customers that want local access more than they want lower price.

        • by LWATCDR ( 28044 )

          I guess it makes sense for a startup that wants access to hardware. Still seems like a large waste of money.

      • The Stephenson Alabama Google data center is where the TVA Widows Creek facility was. It is halfway between Chattanooga and Huntsville. It is ~20 miles from the TN/AL border. It will benefit from TVA water and nuclear power being cheap.

    • Alaska is not exactly ideal for latency.

    • Google has announced the conversion of a TVA coal->steam plant in Stephenson Alabama to a data center. It is on the Tennessee River and they expect it to be a net energy producer. TVA has cheap electricity and they always have the river to aid cooling.

      I am sure there were/are tax breaks but this is a plant scheduled to close in a small town. If Google's energy claims are semi-correct, it is a good move.

    • by tomhath ( 637240 )
      Cooling is done with water, air temperature isn't that important.
  • No jobs though (Score:5, Interesting)

    by hawguy ( 1600213 ) on Tuesday July 14, 2015 @12:11PM (#50109329)

    Too bad datacenters don't bring jobs commensurate with their size and spending. Once the construction is done, it doesn't take many people to run a modern lights-out datacenter.

    • by Anonymous Coward

      I don't know about you but most IT centers are overstaffed with clueless PC monkeys who tell you, "did you try turning it back of and on?" When I typically respond, "so the entire department should reboot their PCs because we can't access the terminal server?"

      • Which is followed by the PC monkey asking "what's a terminal?"

      • by hawguy ( 1600213 )

        I don't know about you but most IT centers are overstaffed with clueless PC monkeys who tell you, "did you try turning it back of and on?" When I typically respond, "so the entire department should reboot their PCs because we can't access the terminal server?"

        Most large datacenters are staffed by a small NOC that coordinates access to the datacenter by service contractors for routine maintenance. The datacenter itself is literally "lights out" most of the time - the NOC uses nightvision cameras to keep an eye on it, but lights are out (usually with motion sensor lights so when you walk through the datacenter, the lighting follows you) There may be a few service techs on staff that go around and replace failed hardware (no hurry since the software routes around d

        • by cusco ( 717999 )

          Add to that two or three dozen guard staff, facilities personnel, a few shipping/receiving people, and lots of contractors to maintain and expand the cable plant, security system, power system, cooling system, UPS, etc. Still fewer people than a WalMart, but good, interesting jobs that are worth having.

      • by mlts ( 1038732 )

        Those days are long since past. Most datacenters tend to have iLO networks, ability to remotely power up/down equipment, extensive monitoring, decent switches, so the only thing SiteOps has to do is plug the cables into the proper ports, and everything is done from remote. The days of walking around with a CD and/or an external hard drive to reimage a PC are long gone... long since replaced by PXE for hardware, or templates and provisioning servers for VMs.

      • I don't know about you but most IT centers are overstaffed with clueless PC monkeys who tell you, "did you try turning it back of and on?" When I typically respond, "so the entire department should reboot their PCs because we can't access the terminal server?"

        The PC monkeys are in an outsourced call center in South Asia. They've probably never seen a server.

        The idealized data center these days is a dark building with a handful of cable monkeys scurrying around like rats in the woodwork. They jack boxes in and out and plug in cables and that's about it. Everything else can be handled from a remote control center.

    • Too bad datacenters don't bring jobs commensurate with their size and spending.

      Prosperity is generated by the efficient creation of goods and services, not by "keeping people busy". Generating value without requiring a lot of work is a GOOD THING.

  • These ginormous data centers are far more efficient per gigabyte of data stored and processed than the typical home computing setup.

    .
    On the other hand, you are effectively donating all your data to some third party knowing that they will be mining it for information to sell, and with little more than a piece of paper to assure you can get it back.

  • Comment removed (Score:5, Informative)

    by account_deleted ( 4530225 ) on Tuesday July 14, 2015 @01:02PM (#50109917)
    Comment removed based on user account deletion
  • Then a lot of people/companies are going to hightail it back to local data centers. It's just a matter of time. Surprised it hasn't happened yet.

    For internet-only companies, the cloud makes a lot of sense, but for a more old-school company it's just too much of a risk that too many are taking.

    Some companies will be forced out of business the first long term cloud stoppage occurs, or even worse the first long term internet outage. Their people cannot do any work at all, cause they shipped it all off to the cloud, then the business cannot function. They are really going to be kicking themselves once they realize that all the money they thought they saved by using the cloud will be the thing that puts them out of business.

    The cloud is still basically a v1.0 product. Too new to base an entire company's operations on until a few massive security breaches or loss of services occur.

    • "The cloud is still basically a v1.0 product." Begging your pardon, but the cloud is really still 0.X.
      • What era are you living in. We're up to cloud v35. v36 comes out next week and v37 the week after. All our users are currently beta testing v35 and will be beta testing v36 next week. Thanks to agile development we will fix bugs on the fly and introduce new ones for users to test on the run. When something breaks we'll blame the user or delete the critical feature in the next release and claim no one was using it.

        That's not to say we don't have a big development strategy, sometime between v40 and v50 we wil

  • Don't they call that "thunder"?
  • So much for saving the world using CFL bulbs.
  • Comment removed based on user account deletion

Think of it! With VLSI we can pack 100 ENIACs in 1 sq. cm.!

Working...