Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror

First Ethernet Switch In Space 141

Posted by CowboyNeal
from the iss-lan-party-arrives dept.
Rebecca will you marry me? writes "The ESA's Columbus laboratory module was added to the International Space Station in February, but Hewlett-Packard has only now chosen to reveal that the LAN onboard Columbus uses a ProCurve 2524 switch." HP admits it was the "most unusual and demanding" project ProCurve has done yet.
This discussion has been archived. No new comments can be posted.

First Ethernet Switch In Space

Comments Filter:
  • Title is misleading (Score:5, Informative)

    by N3TW4LK3R (841526) * on Saturday June 14, 2008 @09:51AM (#23791519)
    From TFA: "Two redundant LAN switches, developed by the European Aeronautic Defence and Space Company (EADS) Astrium, already operate in the ISS network core and now have been joined by HP's ProCurve 2524 switch"

    I sent this in an e-mail to Taco when the article was still in the 'mysterious future' but that message must have been stopped by his spam filter or something.

    Yeah yeah, I must be new here ;)
    • by ggvaidya (747058)
      Did you try the on-duty editor, at - I think - daddypants at slashdot.org? I e-mailed there once, and they fixed the problem that time pretty fast, although of course perhaps my e-mail vanished into the ether and somebody else with the correct e-mail address got in. Still, it's something to do, I suppose.

      I doubt they'll change this, though, "Another switch in space" doesn't have the same ring to it, and neither will the ringing of their cash machine with titles like that.
      • by N3TW4LK3R (841526) *
        Yes, I mailed it to the daddypants address...

        The correct action to take after reading my message would be to prevent the story from reaching the front page, since it's really a non-story if it's not the _first_ switch in space. Maybe it's the cash machine :) or perhaps the e-mail just got lost.
  • by Anonymous Coward on Saturday June 14, 2008 @09:52AM (#23791535)
    they aren't using Linksys routers as well. Password: defaultshuttle
  • I wonder if they'll connect it to the router in space?

    http://www.space.com/spacenews/archive04/ciscoarch_042104.html

  • by Anonymous Coward
    Excellent, that should triple the resale value of my Procurve 2512 switch. Any offers?
  • by cheebie (459397) on Saturday June 14, 2008 @09:54AM (#23791565)
    Is there some reason why a router in orbit would behave differently in any way from a router sitting in a rack in the server room? (Other than floating, etc.)

    • by thomasdz (178114) on Saturday June 14, 2008 @09:58AM (#23791591)
      Extended G-forces during launch might be a good test of how well solder joints, connectors, and other components are made.
    • by Ellis D. Tripp (755736) on Saturday June 14, 2008 @10:00AM (#23791597) Homepage
      Semiconductors generally don't like high-radiation environments, such as outer space. Hence the normal use of specially made high-$$$ "rad-hard" components in space systems.

    • by N3TW4LK3R (841526) * on Saturday June 14, 2008 @10:01AM (#23791611)
      In space, it's exposed to all kinds of radiation that normally gets blocked by the earth's atmosphere.
      This is one of the reasons we try to limit the complexity of electronics sent out to space. (and additionally, shield the hell out of everything)
      I believe the shuttle uses a computer comparable to a 386, for this reason.
    • by Anonymous Coward on Saturday June 14, 2008 @10:15AM (#23791687)
      Heat is not removed from components by airflow because warm air doesn't rise in zero gravity. This means forced convection has to reach more places.
    • by Dogtanian (588974) on Saturday June 14, 2008 @10:19AM (#23791731) Homepage

      Is there some reason why a router in orbit would behave differently in any way from a router sitting in a rack in the server room?
      I suspect there may be some timeout issues due to the network cable connecting them to ground control slightly exceeding the Ethernet spec's maximum length.
    • by muffel (42979) on Saturday June 14, 2008 @10:22AM (#23791747)
      • Cooling: No 'natural' convection
      • G-Forces, Vibration
      • Radiation
      • by whoda (569082)

        • Cooling: No 'natural' convection
        • G-Forces, Vibration
        • Radiation
        Gee, all the same things that the equipment HP designed for the military in the 70's and 80's was required to have.
    • Heat. Things have to run cooler as there's no convection in space.
      • Re: (Score:3, Informative)

        by joggle (594025)
        That's true but probably misleading. They almost certainly have the router in an accessible area so it is exposed to air so it can use convection to cool itself. Ultimately the space station can only get rid of excess heat via radiation but this particular component doesn't need to be designed differently because of thermal issues.
        • by aix tom (902140)
          But there is no convection, since the heated air doesn't move "upward" like it does on earth. It hangs around the heated part without moving off, unless there is some sort of active ventilation.
          • by joggle (594025)
            Sure it convects, it just doesn't move upward. It does move from hot to cold though and form thermal currents. All you need for convection is a gas with sufficient density or a liquid. You don't need gravity.
            • by aix tom (902140)

              Nope. Why should it move, when there is no gravity to make hot air lighter than cold? Why would thermal currents form?

              For example, they did experiments with candles. and they snuffed, because the hot air didn't move away and thus no new oxygen arrived at the flame.

              http://www.straightdope.com/classics/a3_360.html

              • by joggle (594025)
                2nd law of thermodynamics. The entropy of a closed not in equilibrium will increase until it reaches a maximum value. You simply can't keep hot fluids separate from cold fluids, there will be heat transfer and the label we call this type of heat transfer is convection. The problem with the candle is that the rate of oxygen replenishment is too low to maintain combustion. That is not the same as zero convection. A flame has a very high rate of convection and needs a substantial flow of oxygen to maintain the
                • by aix tom (902140)

                  Interesting article, especially :

                  In zero-g environments there can be no buoyancy forces, and thus no natural (free) convection possible....

                  The 2nd law of thermodynamic of course still applies, so the heat will eventually move away from the electronics, but much slower than through convection, where the heated air moves away and is replaced by cooler air. I would imagine it might be equivalent to enclosing the electronics in Styrofoam, which also keeps the air in place.

                  • by joggle (594025)
                    I'd like to point out that your quote from the article references the same straight-dope article you originally linked to. It's true that a candle in a zero-g environment doesn't necessarily maintain combustion but convection does naturally occur, just at a slower rate. However, it would be trivial to add a fan to mitigate this so one way or another they are certainly using convection to cool the electronics.
                    • by aix tom (902140)

                      So to sum it up, we can agree that there is no gravitational convection, however there is some smaller Rayleigh-Benard and/or Benard-Marangoni convection.

                      Which means, to use "off the shelf" equipment they either need to check carefully how much it relies on gravitational convection, or add forced convection by fans or mounting the equipment in a place where additional airflow exist anyway.

                    • by joggle (594025)
                      Yep!
        • by node159 (636992)
          Think you missed the point of this, the issue is that hot air does not 'rise' in space, this has nothing to do with dumping generated heat into space.
          • by joggle (594025)
            I did not miss the point. There are three primary types of heat transfer: radiation, convection and conduction. In space the only method of heat transfer is radiation. Conduction works within solids. However for fluids and gases convection works well with or without gravity. The convection rate will be different depending on the type of fluid and can depend on whether gravity is present, but for a low power device it should be similar with or without gravity.
    • Re: (Score:3, Interesting)

      by rossdee (243626)
      Radiation could be a problem (cosmic rays, solar storms).

      Then there is cooling - even in the ISS you can't use convection since there is no up for the hot air to rise to.
    • Re: (Score:3, Interesting)

      by NMerriam (15122)
      One thing other folks haven't mentioned is the lower atmospheric pressure. I know when we've used laptops in high altitude situations, the LCD displays would sometimes crap out because they are essentially laminates, and separated when the environmental expectation for ambient pressure were not met.

      Of course that specific problem won't affect a switch, but there may be some other unexpected way in which atmospheric pressure is involved. It isn't necessarily easily tested in a high-altitude chamber, since (a
      • Re: (Score:3, Interesting)

        by cyclone96 (129449)
        Actually, the atmospheric pressure of ISS is maintained between 14.1 and 14.9 psi or so, for just that reason. Only the airlocks and some storage volumes are taken below that. Critical equipment is certified to go to low pressures (in case a module depress occurs) but a lot of the non-critical and/or commercial equipment isn't held to that standard (since it could be replaced).

        Apollo flew with about 5 psi of pressure, and Shuttle would sometimes depress the cabin to 10.2 psi in prep for spacewalks, but IS
    • I've ordered equipment I didn't like and had to replace. The ISS doesn't really have such extraordinary environmental requirements as much as the price up screwing up is so much higher. At about $10K per pound, that's about 2000X as expensive as UPS for "shipping and handling".

      http://www.futron.com/pdf/resource_center/white_papers/FutronLaunchCostWP.pdf [futron.com]
    • by Perf (14203)

      The Apollo computer had to deal with high humidity.
      I would think the ISS would have similar issues.
      If not from normal usage, but also resistance to failure when some machine in the room leaks water bubbles.

  • by dreamchaser (49529) on Saturday June 14, 2008 @09:57AM (#23791581) Homepage Journal
    Despite the misleading title (should probably say first OFF THE SHELF switch in orbit), it's pretty cool how they tested these. From TFA:

    Switches from Cisco, D-Link, Avaya, 3Com, NetGear and Hewlett Packard were exposed to extreme levels of radiation in a particle accelerator in Villigen, Switzerland under conditions similar to space.


    Makes one think more about all the radiation crewmembers get exposed to as well, even within the protective embrace of the Earth's magnetic field. That's one of the big hurdles to travel to Mars of course; long term exposure to varying levels of radiation (mostly from the Sun).

    I just think it's geeky-cool that they put them in a particle accellerator for testing though.
  • well.... (Score:5, Funny)

    by KozmoKramer (1117173) on Saturday June 14, 2008 @10:01AM (#23791609)
    I see a job opportunity for a network engineer, or at the very least a network cabling repair guy. Imagine that help desk ticket @ NASA.......
  • by th0mas.sixbit.org (780570) on Saturday June 14, 2008 @10:03AM (#23791623)
    He's a nerd! Save yourself Rebecca!

    (before you mod OT look at submission again ;) )
    • Re: (Score:1, Funny)

      by Anonymous Coward
      This is /. He's a nerd. He needs all the relationship help he can get, and you're scaring her off?
  • by Anonymous Coward
    HP are bottom-of-the-barrel outsourcers now. Trusting HP to provide networking equipment for the space station would be a scaling up of trusting me, an amateur electronics geek, to build radio receivers for emergency workers. I know I can build working kit and I'm fairly cheap, but I've never had to begin contemplating the construction of gear that needs to be so reliable that great efforts will be wasted and people will probably die if I get it wrong. Neither AC's Shack nor HP Procurve switches are designe
    • by cyclone96 (129449) on Saturday June 14, 2008 @02:28PM (#23793637)
      You are absolutely correct, which is why nothing on these networks is something on which life depends.

      I'm going to greatly simplify this, but there are basically three networks onboard the space station. One is mostly off the shelf laptops and networking equipment that runs Windows and is used for crew support (email, procedures, timelines, photos, and such). It frequently needs maintenance, but it does the job. It's also (relatively) easy to certify and plug new hardware into it, so it can be updated frequently as commercial technology advances (for example, later this year the Thinkpad A31p laptops will be swapped out for newer models).

      The second is a payload ethernet network that is used by the payload system to collect and downlink high volume data through the USOS Ku-band system. Failure of this network only impacts science collection and some support activities. These switches are part of this network. The standards are more stringent, but not to the level of stuff on which safety or mission success depends.

      The third network is the core computer system, which is all custom built hardware/software wired up with MIL-STD 1553 data bus. This is the network which runs the core vehicle systems (life support, attitude control, what have you). The hardware and software are developed to a much more rigorous standard than the first two networks (and it obviously costs a lot more and is slower to update because the the long pole of certification and testing). Some of the machines on this network have been chugging along for nearly a decade without failure.
    • by sjf (3790)
      I'm sorry, but isn't this simply racism ? If a product is outsourced (which we know is a euphemism for "made in China") do we have to assume that it is crap ?
      The damn thing sat in a particle accelerator for three years and presumably still worked. Perhaps the Chinese can actually build these things ?

      You point about MIL-SPEC is taken, however, presumably if HP were claiming it was MIL-SPEC they would have done the certification themselves.
      They didn't and NASA did some certification themselves. I don't doubt
  • by Lost Penguin (636359) on Saturday June 14, 2008 @10:09AM (#23791661) Homepage
    In space, no one can hear the NIC scream.
  • by Anonymous Coward
    Do they have a mailing address?
  • by mbone (558574) on Saturday June 14, 2008 @10:18AM (#23791709)
    The Amateur Radio satellites [amsat.org] went to an Ethernet backbone some time ago - over a decade IIRC.
  • by jpellino (202698) on Saturday June 14, 2008 @10:25AM (#23791773)
    ... that was relieved and surprised it wasn't "hub" and "10Base2"?

    • Re: (Score:3, Informative)

      by seanadams.com (463190) *
      ... that was relieved and surprised it wasn't "hub" and "10Base2"?

      I'm not sure if you realize this, but 10base2 (aka thin net) doesn't use hubs. It's a shared 50 ohm coax with tees at each device and terminator plugs on each end. It uses CS/MACD like a hub, but the electronics and physical topology are totally different.
  • Relevent. [securityfocus.com]
  • Cost? (Score:2, Interesting)

    by Straterra (1045994)
    10Mbit switch? Am I the only one who thought "Gee, I would have though NASA could have afforded at least 100Mbit!"

    The only reason I can come up with is the possibility of higher packet loss with all of the radiation. Does anyone know for sure?
  • Why is this news? (Score:3, Interesting)

    by thesandbender (911391) on Saturday June 14, 2008 @10:53AM (#23791943)
    Seriously... zero-g has no effect on this equipment. Yes it has to have more radiation shielding and has to be shock mounted to survive the launch but other that it could be an iPod or a DirectTV DVR. There's nothing innovative about this. They shot an ethernet switch into space... big deal. Call me when someone invents a way to use quantum entanglement to communicate faster than light. That's news.
    • Re: (Score:1, Informative)

      by ahecht (567934)
      Zero-G has a HUGE effect on anything relying on convective cooling. There is no convection in zero-g, so EVERY hot component needs forced air cooling, which you rarely find in a switch.
      • by brxndxn (461473)
        I found a fan in my HP Procurve 2524.. maybe that's how it passed all those amazing tests.
      • by Mattsson (105422)

        EVERY hot component needs forced air cooling, which you rarely find in a switch.
        That's only true for basic edge switches and stuff targeted at home-users.
        Most professional switches rely on fans for cooling.

        • by node159 (636992)
          The problem is that if you place one of these 'passively' cooled switches in space, it will quite merrily cook itself to death as the heat it generates will not dissipate, air is an excellent insulator, you know.
      • Re:Why is this news? (Score:5, Informative)

        by thesandbender (911391) on Saturday June 14, 2008 @12:02PM (#23792397)
        You can't be serious. Almost every enterprise Ethernet switch has fans. Including the terrestial model of the ProCurve 2524.
      • by mbone (558574)
        The ISS has lots of fans. Listen to a TV broadcast from there sometimes.

        In space, the use of heat pipes [wikipedia.org] is also fairly common - I wonder when this technology will start being used with blades in colos, given the density you can rack mount blades.
  • Token Ring? (Score:2, Interesting)

    by aggles (775392)
    Wasn't the first network for the ISS based on Token-ring? I participated in an Ethernet vs Token-ring RFP in the mid-80's against IBM and we lost the bid. We didn't play golf as well.
    • by cyclone96 (129449)
      The Space Station Freedom network design was token-ring, but that design was scratched when the redesign happened in the '90s. The McDonnell-Douglas/IBM avionics team was also dropped in favor of Boeing/Honeywell. The only IBM equipment that was used on ISS were Thinkpads.
  • When I was at Cisco a few years ago, the VP of Corp. Marketing sent out a mysterious e-mail message about Cisco winning a big government contract. The conditions of the project prohibited Cisco from disclosing that they won the contract and Cisco could not publicly mention the "ISS" project.

    The general assumption in the company was that that NASA was using Cisco routers and switches in the International Space Station. I volunteered to be the on-site SE.

    So I doubt that the ProCurve switch is the first et

  • Delay Tolerant Internet [dtnrg.org] (or DTN) is the current version of Vint Cerf's
    "Interplanetary Internet" - basically, making a TCP-like protocol in situations where there may be long delays and no end-to-end connectivity. I thought that there was a test of this on a shuttle flight but cannot find a link, Vint Cerf last year talked [www.cbc.ca] about a test in 2010.

    To me, that is a lot more interesting than just having a switch in LEO.
  • From TFA: The switch was taken straight off the conveyor belt without modification.

    Also from TFA: The switch underwent three years of development, configuration and qualification testing before it journeyed into space.

    Huh?

    • Re: (Score:2, Funny)

      by SBrach (1073190)
      That's what you get for reading TFA.
    • From TFA: The switch was taken straight off the conveyor belt without modification.

      Also from TFA: The switch underwent three years of development, configuration and qualification testing before it journeyed into space.

      Huh?

      I understand they took off-the-shelve hardware, and ran tests on it for 3 years. No hardware modification, no tweaks etc.
  • Hmm wonder if it would with stand an emp blast.
    Tell me the only network up will be hp switches, I'll just kill myself now.
  • It's a nice switch, but for goodness sake. This switch has been obsolete for at least three years!!!

    It's a 24 10/100 port managed switch, with *optional* uplink modules at 1Gbps; fibre and copper available or some propriety stacking modules. It also has a couple of fans!!

    There are far better switches that are passively cooled, use less power, are cheaper and better performing...
    • by mrbooze (49713)

      It's a nice switch, but for goodness sake. This switch has been obsolete for at least three years!!!
      "The switch underwent three years of development, configuration and qualification testing before it journeyed into space."

  • The Russians used a pencil
  • Why?
    These are switches for whom spanning tree is a foreign concept. They claim to support it, but default set up out of the box seems to be if you put 2 crossover cables between 2 procurve switches it will create a switching loop rather than disable one. Not a great idea when our product relies heavily on multicast. It took me a few minutes to figure out why the CPU usage on a workstation that I had just plugged in and hadn't installed anything on was at like 50-60% within about 2 minutes because I wasn't e
    • Maybe they're just capable of configuring the switch before hooking it up to their network? It's not rocket science, after all.

      According to the article, they tested switches from Cisco as well as D-Link, Avaya, 3Com and NetGear. They don't say why the chose the HP over the others though. They did make a mention of the simplicity of the circuits being beneficial, but didn't say whether Cisco's gear didn't survive the tests, or whether HP was simply the cheapest of the surviving devices.

      • by crossmr (957846)
        We worked mainly with cisco switches which came with either stacking modules or had spanning tree enabled by default. There is a reason spanning tree exists and should be generally enabled by default on most switching hardware. I was perfectly capable of enabling spanning tree and did so once I discovered the problem. The problem was simply that it was not enabled by default which is a rather unexpected behaviour for switching hardware at least in my experience and that of my coworkers.

        The only benefit I ev
  • When I worked for state government we were relocating the I.T. unit along with a few other units to a new building. I evaluated Cisco, and HP products and settled on an HP4108 switch. It still works flawlessly.

An inclined plane is a slope up. -- Willard Espy, "An Almanac of Words at Play"

Working...