Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

Exploiting Process Lifetime Distributions for Dynamic Load Balancing

Mor Harchol-Balter and Allen B. Downey

EECS Department
University of California, Berkeley
Technical Report No. UCB/CSD-95-887
November 1995

http://www.eecs.berkeley.edu/Pubs/TechRpts/1995/CSD-95-887.pdf

We measure the distribution of lifetimes for UNIX processes and propose a functional form that fits this distribution well. We use this functional form to derive a policy for preemptive migration, and then use a trace-driven simulator to compare our proposed policy with other preemptive migration policies, and with a non-preemptive load-balancing strategy. We find that, contrary to previous reports, the performance benefits of preemptive migration are significantly greater than those of non-preemptive migration, even when the memory-transfer cost is high. Using a model of migration costs representative of current systems, we find that preemptive migration reduces the mean delay (queueing and migration) by 35% - 50%, compared to non-preemptive migration.


BibTeX citation:

@techreport{Harchol-Balter:CSD-95-887,
    Author = {Harchol-Balter, Mor and Downey, Allen B.},
    Title = {Exploiting Process Lifetime Distributions for Dynamic Load Balancing},
    Institution = {EECS Department, University of California, Berkeley},
    Year = {1995},
    Month = {Nov},
    URL = {http://www.eecs.berkeley.edu/Pubs/TechRpts/1995/5735.html},
    Number = {UCB/CSD-95-887},
    Abstract = {We measure the distribution of lifetimes for UNIX processes and propose a functional form that fits this distribution well. We use this functional form to derive a policy for preemptive migration, and then use a trace-driven simulator to compare our proposed policy with other preemptive migration policies, and with a non-preemptive load-balancing strategy. We find that, contrary to previous reports, the performance benefits of preemptive migration are significantly greater than those of non-preemptive migration, even when the memory-transfer cost is high. Using a model of migration costs representative of current systems, we find that preemptive migration reduces the mean delay (queueing and migration) by 35% - 50%, compared to non-preemptive migration.}
}

EndNote citation:

%0 Report
%A Harchol-Balter, Mor
%A Downey, Allen B.
%T Exploiting Process Lifetime Distributions for Dynamic Load Balancing
%I EECS Department, University of California, Berkeley
%D 1995
%@ UCB/CSD-95-887
%U http://www.eecs.berkeley.edu/Pubs/TechRpts/1995/5735.html
%F Harchol-Balter:CSD-95-887