enormous instabilities @ CA 0800 EDT daily?

Alan Hannan alan at gi.net
Tue Apr 30 17:34:10 UTC 1996


  Et. Al,

] I was looking over some trend graphs which I remember to look
] at from time to time and noticed that there recently has been
] a large level of instability between 7 and 9 a.m. Eastern (D.C.) time,
] with the worst period occurring at about 0800.

  You'll notice a supplemental spike occuring occasionally ~6pm,
  along w/ another around 2am.

] The instability graphs that compute.merit.edu puts together
] shows a very similar trend, at least for the past few days.

  Indeed.

] Is anyone else seeing this?

  It's there, I've never noticed it before, neither in feeling the
  network, or watching graphs......

] If so, does anyone have any theories?  I'd really like to see this fixed.

  1/ Brainless Managers come into work and power cycle routers,
     causing massive updates.

     - no flaws

  2/ Network Engineers are coming into work at this time and
     modifying their routing policies because of visions their dreams
     gave them.

     - flaw:  Real Engineers don't come into work before 10AM, and
              never work past 7am.

  3/ Automated processes by NSP engineers on the West Coast process
     filter modifications and routing policy adjustments at ~4-5AM
     West Coast Time, translating to 7-8AM EST.

     - flaw:  Hmm.

  4/ Initial usage of the internet spikes ~7am, when office
     individuals get to work, causing high utilization, impeding
     routing updates, and causing oddness.

     - flaw:  Routers should be able to handle it... *cough cough*

  None of these are based on empirical evidence, just creative
  thoughts.

  -a



More information about the NANOG mailing list