TWiki> ArdaGrid Web>Dashboard>DemoGuidelines (revision 5)EditAttachPDF

Google Earth

Installation

Installation instructions can be found here.

Steps for demo

  1. Zoom out to show the full globe activity
  2. Explain
    1. Fixed lines are active data links, where GREEN is good status, YELLOW warning status, RED critical status
    2. Blue lines showing up from time to time are processing jobs being picked up by worker nodes
  3. Zoom in: to the site of their choice if they have a specific one, to the Europe region if they don't
  4. Explain
    1. Dots moving along the lines are transfer attempts, where GREEN is a successful one, RED is a failed attempt
  5. Click on a site to pop up additional info
  6. Explain ....

QAOES

QAOES stands for Quick Analysis Of Error Sources. The tool is available via the web interface QAOES.

Steps for demo

  1. load the page and select the 3 input parameters
  2. explain the 3 parameters:
    1. timerange: the input data for the Association Rule mining is job monitoring data taken from the past x hours. Choices are: 6, or 12 hours.
    2. minimum number of jobs, called "minimum support": it is the minimum percentage of jobs in the input data that has to contain all the components (component = attribute-value pair, like site=CERN) in order to generate a rule. The value is given in percent, so between 0 and 100. Choices in the interface are: 0.1%, 0.5%, 1%.
    3. confidence: states how significant the rules have to be. It is the percentage of how many jobs which contain the components of the antecedent also contain the components of the consequent.
  3. click "Show Output" and explain:
    • in general: an association rule is an implication A => C, the left-hand-side is also called antecedent, the right-hand-side consequent. The output of QAOES presents only rules of the format "components" => "Error Code".
    • there are three values for each association rule (line):
      1. support/number of jobs: percentage of jobs and number of jobs in the input data that contain the components and the error code
      2. confidence of the rule
      3. lift: represents the interestingness of a rule. The higher the value, the more "interesting". It is the ratio of the components AND the error occuring together and the components and the error occuring separately, lift(A=>C) = F(A and C)/(F(A)*F(C))
    • for each line, there are a couple of links on the left side:
      1. click on the link to the job summary, which contains the settings of the present components
      2. click on "Add Rule" to add a new rule to the expert system. A rule consists of a problem and a solution. For the problem definition the attributes and values in the present components can be used. The solution is a chain of steps that should be undertaken.
      3. "Show/hide Rules": if there are already rules in the knowledge base which match a present association rule, the rule (with the solution) can be shown/hidden. A rule can be edited and ranked.

SAM

Link

Steps for demo

-- RicardoRocha - 2009-09-15

Edit | Attach | Watch | Print version | History: r7 < r6 < r5 < r4 < r3 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r5 - 2009-09-16 - unknown
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    ArdaGrid All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback