Learn R Programming

mistat (version 0.999)

COMPURESP: Computer Response Time Optimization

Description

The experiment described here was part of an extensive effort to optimize a UNIX operating system.

Usage

data(COMPURESP)

Arguments

source

Pao, Phadke and Sherrerd (1985)

Details

The experiment described here was part of an extensive effort to optimize a UNIX operating system running on a VAX 11-780 machine. The machine had 48 user terminal ports, two remote job entry links, four megabytes of memory, and five disk drives. The typical number of users logged on at a given time was between 20 to 30. 1. Problem Definition. Users complained that the system performance was very poor, especially in the afternoon. The objective of the improvement effort was to both minimize response time and reduce variability in response. 2. Response variable. In order to get an objective measurement of the response time two specific representative commands called standard and trivial were used. The standard command consisted of creating, editing and removing a file. The trivial command was the UNIX system date command. Response times were measured by submitting these commands every 10 minutes and clocking the time taken for the system to complete their execution.

Examples

Run this code
data(COMPURESP)

layout(matrix(1:4, 2, byrow=TRUE))

with(COMPURESP,
     interaction.plot(
       x.factor=F, 
       trace.factor=rep(0, length(F)), 
       response=SN, 
       legend=FALSE,
       type="b",
       pch=15:18,
       ylim=c(-17, -10)))

with(COMPURESP,
     interaction.plot(
       x.factor=B, 
       trace.factor=rep(0, length(B)), 
       response=SN, 
       legend=FALSE,
       type="b",
       pch=15:18,
       ylim=c(-17, -10)))

with(COMPURESP,
     interaction.plot(
       x.factor=C, 
       trace.factor=rep(0, length(C)), 
       response=SN, 
       legend=FALSE,
       type="b",
       pch=15:18,
       ylim=c(-17, -10)))

with(COMPURESP,
     interaction.plot(
       x.factor=D, 
       trace.factor=rep(0, length(D)), 
       response=SN, 
       legend=FALSE,
       type="b",
       pch=15:18,
       ylim=c(-17, -10)))

layout(1)

Run the code above in your browser using DataLab