Process Lifetimes are Not Exponential, more like 1/T: Implications on Dynamic Load Balancing

1994-08-01

Electrical Engineering & Computer Sciences Technical Reports

CSD-94-826

Text

24 p

The Engineering Library

The two decisions defining any dynamic load balancing strategy are:
1. How do we choose a new host (target host) for the migrated process to run on?
2. How do we select which processes to migrate?

Previous load balancing literature has either assumed the process CPU lifetime distribution function is exponential, or has been vague about its properties, and this has greatly influenced the way in which the above two questions have been answered. In this paper we first determine the lifetime distribution function and then explicitly use it in our load balancing strategy.

Our measurements of Unix processes show that the lifetime distribution function varies widely among different workloads, however a rule of thumb (which is far more accurate than assuming an exponential distribution) is
Pr[lifetime > a

Researchers may make free and open use of the UC Berkeley Library’s digitized public domain materials. However, some materials in our online collections may be protected by U.S. copyright law (Title 17, U.S.C.). Use or reproduction of materials protected by copyright beyond that allowed by fair use (Title 17, U.S.C. § 107) requires permission from the copyright owners. The use or reproduction of some materials may also be restricted by terms of University of California gift or purchase agreements, privacy and publicity rights, or trademark law. Responsibility for determining rights status and permissibility of any use or reproduction rests exclusively with the researcher. To learn more or make inquiries, please see our permissions policies (https://www.lib.berkeley.edu/about/permissions-policies).