MUSBUS
Section: User Commands (1)
Index
Return to Main Contents
NAME
MUSBUS - a multi-user performance evaluation tool
SYNOPSIS
run
[ option ]
DESCRIPTION
MUSBUS is designed to support multi-user performance measurements
that are both realistic and comparable between machines
and systems.
The principal performance metric is the time (CPU and elapsed)
required to perform a selected workload as a function
of increasing concurrent load.
A description of the workload profile is one of the
inputs to MUSBUS, and consequently MUSBUS is a benchmarking
tool, rather than a benchmark
per se.
In addition to the multi-user test, MUSBUS supports a battery
of smaller tests designed to measure some aspect of hardware
speed or system implementation efficiency.
These are diagnostic, rather than performance, tests.
Each MUSBUS execution is controlled by a collection
of environment variables and the script
run
that performs error checking, repeated tests for statistical
validity, set up and clean up.
Particular tests may be selected by options in the command line.
FILES
Workload/* workload description
Tmp/* temporary files
Results/* log files
Tools/* ancillary support tools
DIAGNOSTICS
``Benchmark Aborted'' and associated bad news indicating why.
SEE ALSO
Ken J. McDonell:
An Introduction to the Monash Benchmark Suite (MUSBUS),
(included in the MUSBUS distribution).
Ken J. McDonell:
Taking Performance Evaluation Out of the ``Stone Age'',
Proceedings Summer Usenix Technical Conference,
Phoenix, Arizona, June, 1987, pp 407-417.
Index
- NAME
-
- SYNOPSIS
-
- DESCRIPTION
-
- FILES
-
- DIAGNOSTICS
-
- SEE ALSO
-
This document was created by
man2html,
using the manual pages.
Time: 06:14:23 GMT, December 12, 2024