Re: Windoze not rebooted monthly shuts down airports!

From: JF Mezei (jfmezei.spamnot_at_teksavvy.com)
Date: 09/22/04

  • Next message: JF Mezei: "Re: TCP/IP cluster interconnect?" 
    Date: Wed, 22 Sep 2004 00:22:56 -0400

    David Froble wrote:
    > I've also got to disagree with this conclusion. It's not cluster communications
    > that would be the issue. If part of a cluster goes away, fine, as long as all
    > users have access to the part of the cluster that remains available.

    Perhaps Patrick Moreau can chip in.

    I think one has to take a more global look at an ATC application, which
    includes workstations as well as the connection between a workstation and the
    server infrastructure. Availability becomes an interesting issue.

    There is an interesting transactional nature between regions, when a flight
    gets transfered from one regional centre to another. It isn't enough to just
    tell the new centre is it now handling that flight, but there would also be
    collision detection issues to ensure that two flights in adjoining regions
    wouldn't enter a 3rd region on a collision course.

    But inside the region, it isn't so much a transactional system as a data
    acquisition and processing to generate vectors and detect collision etc.

    > With enough shadow sets, or however storage would be set up, all data should
    > remain available to whatever part of the cluster remains available.

    But you are also more likely to see global sections that contain the live data.

    > I'm not real knowledgable about running a cluster, so I'm sure there are others
    > who could provide solutions for the problems.

    Thing is that one should NOT view VMS clusters as the solution to ATC
    problems. We should view the services provided by VMS clustering as tools that
    would allow such an application to be architected in a very elegant way. The
    application would be far more important than the OS in that case.

    > I'm not sure why 96 nodes wouldn't be enough.
    > don't know the requirements so I'm just speculating, but I have to wonder just
    > how many GS1280 systems would be required.

    Whether you have standalone systems or galaxies/wildfires really depends on
    the application architecture.

  • Next message: JF Mezei: "Re: TCP/IP cluster interconnect?"

    Relevant Pages

    • Key Hashing
      ... where cluster is the number of buckets ... that has that number of entries in it. ... First Collision Data ...
      (comp.programming)
    • X-ray satellites discover the biggest collisions in the Universe (Forwarded)
      ... galaxy clusters merging into a giant cluster. ... Undaunted, Renato Dupke and colleagues from the University of Michigan, ... observatories, to disentangle the puzzling galaxy cluster, Abell 576. ... Dupke realised that Abell 576 is also a collision, but seen head on, so ...
      (sci.astro)
    • X-ray satellites discover the biggest collisions in the Universe (Forwarded)
      ... evidence that galaxy clusters can collide faster than previously thought. ... Undaunted, Renato Dupke and colleagues from the University of Michigan, Ann ... observatories, to disentangle the puzzling galaxy cluster, Abell 576. ... Dupke realised that Abell 576 is also a collision, but seen head on, so one ...
      (sci.space.news)
    • Galaxy Cluster Collision Left a Clumpy Aftermath
      ... Pandora's Cluster probably formed from the collision of four different parent clusters, a process that lasted millions of years. ... When two of the parent clusters smashed together, their collision created a central "bullet," where the impact of gas on gas created a shock wave in matter but left the distribution of dark matter untouched. ...
      (sci.astro.amateur)
    • Galaxy Cluster Collision Left a Clumpy Aftermath
      ... Pandora's Cluster probably formed from the collision of four different parent clusters, a process that lasted millions of years. ... When two of the parent clusters smashed together, their collision created a central "bullet," where the impact of gas on gas created a shock wave in matter but left the distribution of dark matter untouched. ...
      (sci.physics)