|
Defect Density Prediction
Software reliability assessments
Software reliability support
Software Root Cause Analysis
Gap analysis
| |
Software reliability assessment and predictions
- Are your customers requiring you to address software reliability during the proposal
phase?
- Do you need to address software reliability during the proposal phase to be competitive?
- Have your customers established software reliability objectives?
- Do you have suppliers that are in need of guidelines for delivering reliable software?
- Are you in the process of writing a statement of work pertaining to the reliability of
delivered software to your organization?
If any of these are true, we can work with you to address software reliability measurement
requirements during the pre-proposal, proposal, concept and development phases. The
techniques available for measurement are:
The software reliability assessment and prediction process
The below figure illustrates the process for assessing and predicting
software reliability.
First a survey is completed. This is the Software
Reliability Assessment step. The result of the survey is a predicted
percentile group
Second is the measurement step.
- The predicted percentile group is used to predict
defect density and probability of the project being late. The normalized
effective size is predicted and multiplied by this to yield the number of
fielded defects.
- As an optional part of this step, each of the survey answers can be
benchmarked line by line to the the answers to
the same questions from other organizations in the same or similar industry.
- Next the predicted fielded defects is then transformed into a predicted MTTF, MTBCF,
failure rate, reliability, availability.
- Once the software is in a testable state the reliability can be
estimated using reliability growth models
Third is the analysis step. The software can also be analyzed
using a root cause analysis, SFMEA or
SFTA.
Fourth is the improvement step. The predicted percentile group can be used to determine
quantitative improvements in reliability via the reduction of defect density.
- Strengths and gaps can be
determined directly from the negative answers on the survey.
- The gaps are analyzed to determine which ones can be addressed in the
immediate timeframe. Some gaps are more
sensitive to defect density than others. Some gaps have prerequisites in
order to mitigate.
- Some gaps are more relevant to some industries than others. Some gap mitigations will impact schedule or budget
more than others. Some gaps are inexpensive but require significant
culture change. Some gaps require alot of training to address. The model can be used to identify a set of improvements
that balances all of these constraints.
The below illustrates the parts of the prediction process and the associated
tools/services that SoftRel, LLC provides
Key Benefits
- Predict the defect levels, MTTF, failure rate and reliability before the coding and
testing begins
- Use the results to drive development practices to achieve
a specific goal instead of being surprised by an unacceptable reliability
level later when it's more expensive to address
- Use the results to determine the key development practices that have the highest return
on investment
- Determine person power needed to support the desired or required reliability
requirements
- A completed report and/or proposal for your customer
- This method is also automated in Frestimate
Capabilities
The owner of SoftRel, LLC has 20 years of experience in measuring software reliability and
in developing proposals or statement of work documents specifying software reliability
objectives.
Deliverables
- Includes software reliability assessment.
- Predicted defects and defect density at start and end of testing as well
as any time during useful life
- Key milestones are start of test, end of test, average during useful life,
end of useful life (point at which software version is replaced with a new
major software version
- Predicted failure rate, critical failure rate, MTTCF, MTTF, availability and reliability
- Predicted staff required to fix residual defects each month after delivery
- Predicted staff required to test the software to reach a desired level of defect density
at delivery
- Determine how software failure rates and components will be merged into the overall
system reliability block diagram
- All of the above predictions are completed for 4 milestones
- Start of test
- End of test (deployment)
- Average useful life
- End of useful life (point at which software version is replaced with
next major software version)
Optional services
| Service |
Description |
Notes |
| Gap analysis performed at SoftRel,
LLC
|
- Identification of key gaps that separate your project from getting to the next
percentile
- Identification of key strengths that separate your project from percentiles with bigger
defect density
- Recommendations as applicable for how to execute the key gaps identified
- Checklists
- Helpful tips and hints
- Lessons learned
|
You must have an initial and detailed assessment prior to
having a gaps and strengths analysis |
| Software reliability program plan |
See a sample table of contents |
An initial and detailed assessment is recommended but not
required |
| Get benchmarked |
See get benchmarked |
An initial assessment is required. A detailed
assessment is highly recommended. |
| Root cause analysis |
Before you make any improvements you should know the most
common root causes. This analysis is also a prerequisite for a SFMEA. |
This can be done without a software reliability assessment |
| Software FMEA service |
Testing often focuses on the success scenarios. A
SFMEA will focus on the failure scenarios. We can
train you to perform the SFMEA or we can perform
the service for you. |
Root cause analysis |
| Software FTA service |
Software is often ignored when doing a system fault tree.
Learn how to include software on the FTA. |
This can be done without a software reliability assessment. |
Copyright, SoftRel, LLC 2011. This page may not be
copied in part or in whole without written permission from
Ann Marie Neufelder
|
