Hardware and
Software Reliability (323-08)
Application and Improvement of
Software Reliability Models
Submitted By: Dolores Wallace
And
Charles Coleman, Manager, SATC
October 12, 2001
|
Technical
POC: Dr.
Linda Rosenberg |
Administrative
POC:
Dennis Brennan |
|
|
|
|
Phone #: 301-286-0087 |
Phone #: 301-286-6582 |
|
Fax #: 301-286-1701 |
Fax #: 301-286-1667 |
|
Email: Dennis.Brennan@gsfc.nasa.gov |
|
|
Mail Code: 304 |
Mail Code: 300 |
This report presents the results of Task 323-08, Hardware and Software Reliability. Although hardware and software differ, they share a sufficient number of similarities that the mathematics used in hardware reliability modeling have been applied to software reliability modeling. This task examines those models and describes how they may be practical for application to projects at Goddard Space Flight Center. The task also resulted in improvements to one model to allow for fault correction.
NASA acquires and uses many systems in which
software is a major component, and many of these systems are critical to the
success of NASA’s mission. These
systems must execute successfully for a specified time under specified
conditions, that is, they must be reliable.
The capability to provide accurate measurement of the reliability of the
software in these systems before NASA accepts them is an essential part of
ensuring that NASA software will meet its mission requirements.
The purposes of Task 323-08, Hardware and Software
Reliability, are to examine reliability engineering in general and its impact
on software reliability measurement, to develop improvements to existing
software reliability modeling, and to identify the potential usefulness of this
technique as one data point in measuring reliability of software at the Goddard
Space Flight Center.
The first part of this project identified the
mathematics and statistical distributions used in reliability modeling and
found that essentially all have been applied to software reliability modeling.
The study identified major differences between hardware and software and indicated
that the software reliability models do not specifically accommodate those
differences. The study resulted in
several recommendations for model modification.
The second part of this project explored the use of
these software reliability models at Goddard Space Flight Center (GSFC) and
their improvement. A case study to determine usefulness of this technique at
GSFC used project failure data and characterized and manipulated it for use
with a software reliability tool. The
actual process of software reliability modeling includes the preparation of the
data, selection of the appropriate model, and analysis and interpretation of
results of the models. A key criterion
to practicality is the amount of effort required for each step.
The Naval Space and Warfare Center (NSWC), Dahlgren,
Virginia, has sponsored the development of a software tool, Software Modeling
and Estimation of Reliability Functions for Software (SMERFS), under the
direction of Dr. William B. Farr. This public domain tool exercises several
software reliability models and served as an instrument for assessing usability
of software reliability modeling at GSFC
One difference between hardware and software is the
correction process. By the time
hardware is in operation and reliability studies occur, generally design faults
have been removed. With software, faults are often remaining during system test
and operation. The hardware reliability models do not account for correction
during the time of reliability measurement. For this research task, Dr. Norman
Schneidewind of the Naval Postgraduate School developed adjustments to the
Schneidewind model to allow for fault correction.
This report describes the results of these studies.
2. Overview of Hardware and Software
Reliability
2.2 Software reliability
models
2.3 Requirements for using
the models
2.4 Some Hardware and Software Differences Impacting Reliability Models
3. Applying Software Reliability Modeling at GSFC
3.3 Software Tool Availability
3.4 Options for applying
software reliability modeling at GSFC..
4. Modeling the Fault Correction Process