18
EGEE-III INFSO-RI-222667 Enabling Grids for E-sciencE www.eu-egee.org EGEE and gLite are registered trademarks Modeling Grid Job Time Properties Lovro Ilijašić Lorenza Saitta University of Eastern Piedmont, Italy

Modeling Grid Job Time Properties

  • Upload
    mabli

  • View
    26

  • Download
    1

Embed Size (px)

DESCRIPTION

Modeling Grid Job Time Properties. Lovro Ilijašić Lorenza Saitta University of Eastern Piedmont, Italy. Grid Observatory. The Grid Observatory cluster of EGEE – the scientific view Data collection, analysis of behaviour and usage 20 months of data, more than 28 million jobs - PowerPoint PPT Presentation

Citation preview

Page 1: Modeling Grid Job Time Properties

EGEE-III INFSO-RI-222667

Enabling Grids for E-sciencE

www.eu-egee.org

EGEE and gLite are registered trademarks

Modeling Grid JobTime PropertiesLovro Ilijašić

Lorenza Saitta

University of Eastern Piedmont, Italy

Page 2: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667 Modeling Grid Job Time Properties 2

Grid Observatory

• The Grid Observatory cluster of EGEE – the scientific view

• Data collection, analysis of behaviour and usage• 20 months of data, more than 28 million jobs• Development of models

• Grid is more than just a sum of its parts

Page 3: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Emergent Behaviour

• Properties that are apparent only on higher levels of organization and are not present on the lower ones

• Emergent Behaviour is observable on all levels of reality

Modeling Grid Job Time Properties 3

Page 4: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Power Law

Pareto distribution Zipf’s law 80-20 rule Self similarity

Modeling Grid Job Time Properties 4

Page 5: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Degree Distributions

• In- and out-degree distributions: How users connect (use) CEs

• Weighted degrees: Distribution of number of jobs

Modeling Grid Job Time Properties 5

Page 6: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Job Lifecycle Analysis

Modeling Grid Job Time Properties 6

Page 7: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Distributions of Job Lengths

Modeling Grid Job Time Properties 7

Page 8: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Distributions in log-log scale

Modeling Grid Job Time Properties 8

Page 9: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Power-law vs. Log-normal

• Power-law: preferential attachment• Power-law: optimization of the average amount of

information per unit transmission cost• Power-law: monkeys typing randomly• Probabilities of letters not equal: power-law or log-

normal?

Modeling Grid Job Time Properties 9

Page 10: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Log-normal vs. Power-law

• Log-normal: multiplicative processes

• At each step, the event (Xt) may grow or shrink, according to a random variable Ft: Xt = Ft Xt-1

• Multiplicative models can also generate Pareto distribution if there is not a minimum size of event. Otherwise it is log-normal

• Intermixing of generations, where t is random variable, leads to power law.

Modeling Grid Job Time Properties 10

Page 11: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Log-normal Fitted Distributions

Modeling Grid Job Time Properties 11

Page 12: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Alternatives

• Double Pareto distribution• Double Pareto log-normal distribution• More distribution parameters that allow better fitting

Modeling Grid Job Time Properties 12

Page 13: Modeling Grid Job Time Properties

EGEE-III INFSO-RI-222667

Enabling Grids for E-sciencE

www.eu-egee.org

EGEE and gLite are registered trademarks

Modeling Grid JobTime PropertiesLovro Ilijašić

Lorenza Saitta

University of Eastern Piedmont, Italy

Page 14: Modeling Grid Job Time Properties

Modeling Grid Job Time Properties 14

Page 15: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Complex Networks

• Complex Networks – Complex systems represented as graphs

• Gathered experiences from Physics, Chemistry, Biology, Computer Science, Sociology, Economics…

• Representing Grid as a Complex Network

• 20 months of log data, more than 28 million jobs

• Edges representing jobs go from Users to CEs

Modeling Grid Job Time Properties 15

Page 16: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667

Number of jobs for each user

Modeling Grid Job Time Properties 16

Page 17: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667 Modeling Grid Job Time Properties 17

Page 18: Modeling Grid Job Time Properties

Enabling Grids for E-sciencE

EGEE-III INFSO-RI-222667 Modeling Grid Job Time Properties 18