Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Microsoft Aims To Cure Server-Hugging Engineers 285

1sockchuck writes "Microsoft wants the engineers in its labs to manage their servers remotely, and is moving development servers from a bevy of computer rooms in labs to a new green data center about 8 miles from its Redmond campus. 'I see today as a real transition point in our culture,' said Rob Bernard, chief environmental strategist at Microsoft, who acknowledged that the change will be an adjustment for veteran developers but will save money and energy use. Microsoft expects its customers will run their apps remotely in data centers, and clearly expects the same of its employees."
This discussion has been archived. No new comments can be posted.

Microsoft Aims To Cure Server-Hugging Engineers

Comments Filter:
  • lol. (Score:5, Insightful)

    by Anonymous Coward on Wednesday September 09, 2009 @01:39PM (#29368471)

    welcome to 1970s unix, M$.

    • by Anonymous Coward

      This is all about outsourcing.

      The more they can develop and use technologies where engineers don't need to be near the machines, the easier it is to fire everyone in western countries and hire people from India, etc...

      Not that this necessarily was a bad thing. They do need the work just as much (or well, probably a lot more) than we do.

      • Couldn't they just locate some dev servers in India, like companies have been doing for over a decade now?

        There will always be some development work that must be done locally, and some can be done from across the world just as easily. It all depends on the task and how much business interaction is required.

      • I knew I should have been a programmer. As each year goes by my hardware skills become less-and-less useful, because they keep moving the equipment to remote locations and handing-over control to programmers or administrators (or lawyers).

        Oh well. (signs up for college). Time to earn that second degree (and maybe score with the ladies for a change).

        • by Reapman ( 740286 )

          As someone who is getting into development, I wouldn't say it's a great career choice. I don't think we're any safer from the outsourcing then the hardware guys. I'm in it because i love it, but otherwise I'd get out.

        • As a developer, I've often thought the inverse.

          That being said, I think the outsourcing fever has largely run its course in development. More managers have come to learn the hard way that some development can be smart to outsource, but it's a lot less than the "nearly everything" than they thought five years ago.

        • I knew I should have been a programmer. [snip] Time to earn that second degree (and maybe score with the ladies for a change).

          Yep, those babes sure do go for the code geeks!

      • Re: (Score:3, Insightful)

        by elrous0 ( 869638 ) *
        MS better hope not. If everyone in the West is out of work, who'll be left to buy legitimate copies of Windows and Office?
      • The Asians are getting to expensive. That's why we need to build up Africa's infrastructure so we can outsource to them instead. Arguably it's nice to send work to the people that need it the most but it's also abusing people by shifting work away from people just to save a quick buck.

        Why don't companies build better products to increase profits? Software in general is in dire need of better quality.
        • by dfetter ( 2035 )

          Because better products don't increase profits per se. Getting paid more than you spend increases your profits.

      • They do need the work just as much (or well, probably a lot more) than we do.

        But that doesn't mean that they should steal our jobs to get it. No reason why they can't develop things in their own country instead of taking our stuff. (how many Americans go and take stuff from India?) Mod me down if you wish, but the truth is that outsourcing only benefits shareholders and upper management. How long do you will it be until the Indian workers who got our jobs get cast aside by the same companies when someone

      • by Venik ( 915777 ) on Wednesday September 09, 2009 @04:05PM (#29370761)
        It is about outsourcing. Remote support is not a bad thing in itself. I've been supporting most of my systems remotely for almost four years now. But when there is a hardware problem, I can be at the site in under an hour. Our operators are also remote and are not much help. Those of you with experience supporting Linux clusters know that hardware problems are much more common with these systems than with enterprise-class Unix servers. I keep a record of all the failed parts and let them pile up for a couple of weeks, so I don't have to drive to work every day. We do have a designated touch labor guy for onsite support, but he was dropped on his head when he was a child. Aside from keeping the servers running, my other duty is to throw sand in our management's dreams of outsourcing everything to India. I have nothing against Indians - they need jobs like everybody else. However, most computer specialists from India I had a chance to work with over the years were grossly incompetent. Perhaps it's just my bad luck. Or, maybe, it is very silly of us to expect a country with one of the lowest numbers of computers per capita in the world to produce so many qualified computer engineers.
  • Wait what? (Score:3, Insightful)

    by Monkeedude1212 ( 1560403 ) on Wednesday September 09, 2009 @01:42PM (#29368525) Journal

    I've been remoting into my servers for years, because Microsoft Active Directory and some DNS services makes it so easy...

    And Microsoft hasn't?

    • Re:Wait what? (Score:5, Insightful)

      by Culture20 ( 968837 ) on Wednesday September 09, 2009 @01:51PM (#29368669)
      I remote into my servers too, but do you really want to drive eight miles away to diagnose a potential hardware issue, or relinquish physical control to a dedicated hardware monkey?
      • Re:Wait what? (Score:4, Informative)

        by L0rdJedi ( 65690 ) on Wednesday September 09, 2009 @02:01PM (#29368827)

        I remote into my servers too, but do you really want to drive eight miles away to diagnose a potential hardware issue, or relinquish physical control to a dedicated hardware monkey?

        I already do if I'm working on a server at night and it becomes unresponsive or fails to reboot properly. Work is about 7 miles from home. If the server isn't back up within 10 mins of a restart, it's off to the office to figure out why.

        During working hours the only time I need physical access to anything is when I'm changing the backup tapes...that's once a week. I do everything through RDP and VNC.

        • Re:Wait what? (Score:5, Insightful)

          by PainKilleR-CE ( 597083 ) on Wednesday September 09, 2009 @02:13PM (#29368979)

          Most of my servers are ~10 floors over my head and I still have to call someone to let me into the room if I need physical access to them. My production servers are in another state and I doubt anyone on my team has ever seen them. There's a lot to be said about having physical control over the hardware when you want it, but there's also a lot to be said about making it someone else's job to make sure you don't need it. It also teaches you a more proactive approach to server management.

        • In a virtualized server environment, the problem of keeping the hypervisors up isn't yours. The problem of keeping the virtual servers running, may or may not be yours, but if it is, you will have a control client access layer that is akin to access to the HW.


          • Re: (Score:3, Informative)

            by afidel ( 530433 )
            Yeah except a lot of those engineers are going to be developing the hypervisor for the next generation of Windows Server. Of course then you just use iLo/DRAC/ASM to remote in or use an IP KVM and a port addressable PDU to power cycle the hardware. Heck I'm the guy responsible for the datacenter at my employer and I rarely go into it, it's too noisy and either too hot or too cold depending on which side of the aisle you're on. My DR equipment is colo'd with AT&T several states away and we've needed remo
      • Re: (Score:2, Insightful)

        by fredjh ( 1602699 )

        Yup... I'm not really an administrator, but I have a couple of departmental web servers I need to run... and damnit, sometimes I gotta hit the button.

      • by siloko ( 1133863 )

        but do you really want to drive eight miles away to diagnose a potential hardware issue

        I get around this by keeping the hardware close at hand but running my servers in VM's - where they reside is anyones guess!

      • Re: (Score:2, Insightful)

        by DogDude ( 805747 )
        "relinquish physical control to a dedicated hardware monkey?" Why wouldn't you want to do that? Why would a company want to pay an admin salary for somebody to swap out hard drives?
      • If you work in a large company, you might already have that situation. Because the server room is considered too sensitive to let everybody who feels qualified tinker with the servers. My last job was in a medium sized company in the 100-200 employee range, and even there the server room was locked up. With access only for the dedicated hardware monkeys.

      • Server grade equipment shouldn't be having hardware issues anymore not detected by the management module. In something like a POWER BladeCenter if you have power to the chassis at all, you can do nearly everything remotely you can do standing in the room. Unless it has a physical problem it'll run forever like that.

      • Potential Hardware issue?

        In my experience hardware works or does not. Generally if a server is always on, then when it fails if completely fails. If it has failed it would probably take about a day to figure out what went wrong and replace it. This is a strange coincidence, because for one day of my time you can buy a new server.

        When servers fail, they are usually now thrown in the bin since figuring out what failed is not worth the time a skilled employee will spend on it. Hardware is now so cheap, that di

    • by rrhal ( 88665 )
      Microsoft IT has been for years. They need a tech on location to go recover from BSOD and troubleshoot SAN issues. Otherwise they are run from the main campus in Redmond or increasingly from Hyderabad.
    • This isn't about IT, which is already remotely administered in datacenters around the world. Product groups generally have their own labs containing servers for running builds, tests, etc. These are also remotely administered, including things like installing OS builds. The new datacenter is a way to consolidate these small labs.

  • Good idea (Score:5, Funny)

    by Anonymous Coward on Wednesday September 09, 2009 @01:43PM (#29368551)

    I want to be as far away from Windows as possible.

  • 1985 called (Score:5, Funny)

    by convolvatron ( 176505 ) on Wednesday September 09, 2009 @01:44PM (#29368555)

    they want their tty back

    • Well, in their defense, they were a bit late, but they eventually released real mode Windows for Workgroups (at least for those of us who had NetBEUI installed), didn't they? Oh, wait ...

      I want my life back Bill Gates!

    • Re: (Score:3, Funny)

      I read this as "they want their titty back". Oh my. The net has effectively turned me into a female body addict. http://domai.com/ [domai.com] Ooops that slipped right out.

  • by sitarlo ( 792966 ) on Wednesday September 09, 2009 @01:45PM (#29368573)
    How can a concrete, environmentally controlled, power sucking, place people drive to be considered green? Oh, I know, they're using 36 Watt light bulbs in the broom closets.
  • by Culture20 ( 968837 ) on Wednesday September 09, 2009 @01:46PM (#29368587)
    By installing some sshd by default (openssh comes to mind)? Pretty please with gumdrops? All the other operating systems are doing it. As nice as they are, psexec et al are just not the same.
    • by CAIMLAS ( 41445 ) on Wednesday September 09, 2009 @02:23PM (#29369091) Homepage

      Unfortunately, not exactly. MS will just use RDP. It's a decent enough protocol on its own, and better than many others in its domain.

      On the downside for MS, this move will likely mean a bigger focus to find and exploit holes in RDP. Until now, I don't think there have been many (in no small part because RDP has been relegated to internal terminal server use and remote in-house networking - Windows admins don't seem to like it all that much, at least compared to *nix admins who love SSH).

    • by sorak ( 246725 )

      If they did that, it would be hard for them to justify charging per seat for an MSTSC license.

  • A losing battle (Score:2, Insightful)

    by sskinnider ( 1069312 )
    I have been through three "server consolidation" projects in the last 13 years. Every time the IT shop moves servers from the wild into data centers, a whole new crop springs up in wild. Good policy and great enforcement are the keys to a successful consolidation, but with the do more with less mentality of most IT shops today the old just becomes the new again.
    • by zaanan ( 1617787 )
      The customer hammers on the IT department to cut costs, so the IT dept. consolidates their servers, ostensibly to reduce support overhead. However, some opportunity costs of consolidation (more security & controls requirements, lengthy endorsement/approval process, etc.) make it an onerous process to put servers into the datacenter, so then local "server rooms" start springing up again, especially for testing & development. Plus there are some things (security video, etc.) that you need to have loca
    • by BuR4N ( 512430 )

      a whole new crop springs up in wild

      The trick is to switch the fuses after the server move so they barley can run a workstation on each outlet.

      • Sounds like it would cause a lot of lost work as those breakers on the edge of tripping decided they were going to trip (due to some small surge or whatever)

        Also in many places computers aren't the only things that need to be plugged in.

        Personally I think development servers are fine under the devs desks. Easy for the devs to deal with if they fuck something up. Low density so no real cooling problem and if the power goes out in the office the dev servers wouldn't be much use anyway.

    • There's servers in the wild, and then there's this guy's problem [thedailywtf.com].
  • by Rich0 ( 548339 ) on Wednesday September 09, 2009 @01:53PM (#29368703) Homepage

    This sort of thing is a perfect application for VMWare. Create some interface where engineers can order up a server, and poof, a cloned vmware system is provided to them. Then they can have console-level access to that single server and do whatever they want with it. When they're done they hit a button, and poof it is disposed of. Since these kinds of development systems tend to sit around idle most of the time you can oversubscribe the hardware.

    If you must use physical servers then there are lots of remote administration options. Of course good old RDP works just fine for 95% of the tasks. If you're actually working on OS-level changes then you might need a way to remotely boot off of CDs and get remote console-level access. Lots of server-grade solutions provide this kind of capability. VMWare does as well.

    • You are right, VM's are awesome fro QA and Development work. The problem is, how do you keep track of how many VM's are running what? I know its not a problem in the real world. But at your office, do you KNOW that you have licensing for every copy of SQL server, or CALs for windows, etc. Moving things to a virtual environment at my last company was the main driver to start moving things to free/open source, because the time needed to track licensing compliance for all tools and software used got expone
      • where "in the real world" = At Microsoft.. Opps
      • Assuming the machines are only used for development purposes then one option is to get MSDN subscriptions for all the developers who will use the group of dev servers. MSDN subscriptions are licensed by user not by install.

    • by rbanffy ( 584143 )

      Yeah. Right. Seriously. Do you really think Microsoft will become a huge case for VMWare?

    • You mean like this?

      http://src.enomaly.com/ [enomaly.com]

  • They will have to get used to remote handling of BSOD. They will need to have humans local to the actual machines to press the reset button when a BSOD occurs. A new breakthrough will then be announced called "proximity computing" whereby they will move the computers near those using them to avoid the extra employees who were hired to press the reset button. Buy shares in companies that are used to move the computers backwards and forwards each time the management reverses their previous "breakthrough".
    • by faedle ( 114018 )

      The reality of the situation is one $15/hour NOC monkey can effectively be "hands on" support for hundreds.. if not thousands.. of machines. Concentrating machines in one place has a lot of advantages of scale, including security, power, and cooling savings. The costs saved can easily pay for 24 hours of cheap "hands on."

    • by Shads ( 4567 ) <shadus@@@shadus...org> on Wednesday September 09, 2009 @02:37PM (#29369339) Homepage Journal

      In the modern era we use this thing called a "remote access card" or an ip enabled pdu.

      If you're at the datacenter and you're not:

      A) Installing or Removing hardware
      B) Physically rewiring something
      C) Replacing a failed piece of hardware

      You're doing it wrong.

      • Indeed. I am surprised at the amount of comments about booting BSOD etc. I thought that people here would about remote control.

        It's not rocket science anymore. I have managed about 100 servers on a remote location for a couple of years and only had to visit them once or twice every month. If I had to install new servers, I would just go there and install the hardware and then head back to the office for the software installation.
        Most of the time they should just have a standard image installed but when I ne

    • Because you can't reboot a server with an ILO?
    • I honestly haven't used it, but it came up [42u.com] over a project a few years back and I'd been meaning to try it out myself. Our data center is about 5 miles from the office and about 15 miles from my home and we manage mostly via ssh/rdp/vnc with the occasional on-site visit as needed.
    • Sounds like you need an ITAPPMONROBOT [thedailywtf.com]
  • by MosesJones ( 55544 ) on Wednesday September 09, 2009 @01:56PM (#29368735) Homepage

    One thing you can always remain impressed by Microsoft is how they manage to spin something that everyone has been doing for 20 years and talk about it as a trend.

    SMEs are using Rackspace and the like, people are shifting stuff to Amazon Web Services and Microsoft's own strategy is about Azure and the cloud with virtualisation as "normal". In other words what Microsoft are doing here is well behind what they are talking about in the market as being normal.

    But they've still managed to spin a press release out of shifting a bunch of servers into a Data Centre in the sort of move that wouldn't have got any press coverage 10 years ago. Brilliantly however they've added a "green" angle to it all thus turning what looks like a move they should have done ages ago into something worthy of comment.


    You have to admire a press release in 2009 that can make shifting to a DC sound like a revolution.

  • Expects their customers to run Visual Sourcesafe, I hope they expect the same from their engineers.

  • In our data center we've moved a lot of the Windows workloads to virtual servers. That gives you "the console" and "the hardware" even when you're remote, so it's very nice.

    Of course, Microsoft will want to run its virtualization on Windows so that could pose a problem for them. :)
    • Re: (Score:2, Funny)

      by arndawg ( 1468629 )

      In our data center we've moved a lot of the Windows workloads to virtual servers. That gives you "the console" and "the hardware" even when you're remote, so it's very nice.

      IP KVM called. They want to know what year it is!

  • by Hurricane78 ( 562437 ) <deleted@slashdot ... minus physicist> on Wednesday September 09, 2009 @02:15PM (#29368995)

    How do you cure engineers from loving their servers?

    By installing Windows on them! *BA-DUM-TISH*

    I'm here all night! Try the cake!

  • What I want in a remote control box:
    * alternative boot media in case HD won't boot, e.g. bootable CD in the drive
    * remote access to keyboard, video, and mouse
    * remote access to power switch

    All of these are available today. Now for my final requirements:

    * Cheap
    * Secure - only authorized users can get remote access


    • Re: (Score:2, Informative)

      by FaxeTheCat ( 1394763 )
      In our company we do all that on our HP servers. Including AD authentication to the iLO interfaces... well you may not find it cheap, but it is worth every $$$ (especially when you are in Norway and the server is in Nigeria...)
  • by sniperu ( 585466 )
    For all the people who are wondering how do you remotely reboot a server once it BSOD. In the last century or so, there have been these little things called remote cards, or out-of-band management cards, or ILO (HP) or RSA (IBM) or whatever. You can do all kinds of magic with them cards, like remote reboots, connecting remotely to the servers mouse/keyboard/screen, hardware diagnostics, turning on the little light on the server so you can find it once you get to the data room.... Makes me wonder about the
  • by leromarinvit ( 1462031 ) on Wednesday September 09, 2009 @03:07PM (#29369845)
    The headline just begs to recycle this [slashdot.org] old post...

    Actually, I worked with one fellow who had his penis injured by a computer.

    Some of IBM's mid-range systems from the late 1980s (actually quite large, physically, by today's standards...) had a circular opening about 2 inches in diameter. This opening was near some circuitry or device that would heat up rather quickly. So with the help of some duct tape and foam, this hardware admin fashioned himself a warm vagina of sorts, right on the side of our IBM system.

    We're not sure how long he had a "relationship" with the system, but it came to an end one day when during lunch he ran over to a group of us, with his hands covered in blood. Apparently the foam vagina tore, and a piece of metal got him on the penis shaft. He went to the hospital, and was okay in the end. But he didn't really last long with the company after that...

  • This will work out great, I'm sure -- until someone manages to BSoD the server they're working on, and need to physically go and reboot it.
    • Re: (Score:3, Informative)

      by argent ( 18001 )

      Remote controlled power strips, remote consoles (like RealWeasel [realweasel.com] or HP iLO), and so on mean remote PC management is almost as convenient as remote server management. It's not serial consoles, but it's workable.

  • I bet the OS group gets to keep their hardware. While windbg via named pipe is cute, the last thing you need during OS/driver bringup is more complexity. I tend to do just find with IPMI/ILO/IP-PDU/RS-232/etc for remote management during kernel work, but there are some weeks where I have to walk into the lab 100 times a day to move a card/cable, or press the reset button because the particular machine isn't on the IP-PDU and the buggy IPMI card got confused due to a PCIe bus hang or whatnot.

  • ... how sysadmins and managers feel about taking a trip to replace faulty hardware, upgrade and test kernel modules and drivers...
  • "Microsoft expects its customers will run their apps remotely in data centers, and clearly expects the same of its employees." Apparently Microsoft cannot distinguish between the functions of a customer and their engineers.

It is not for me to attempt to fathom the inscrutable workings of Providence. -- The Earl of Birkenhead