SRC Forum - Message Replies
Forum: Reliability & Maintainability Questions and Answers
Topic: Reliability & Maintainability Questions and Answers
Topic Posted by: Reliability & Maintainability Forum
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998
Posted by: Xiangping,Wu
Organization:SHENZHEN HUAWEI TECH.CO.LTD.
Date posted: Thu Sep 16 2:51:17 US/Eastern 1999
Subject: Software Reliability Prediction
I have just bought the book of Reliability Toolkit:Commercial Practices Edition from RAC.I have some questions on the method of software reliability prediction in it.First,in this book,only seven programming languages'expansion ratios are provided.Does anyone could give me the expansion ratios of other programming languages(e.g.VB,VC++,DELPHI,etc.)? The second question is I found when predict a software' initial failure rate by this method,the result is only related to the host processor speed and the programming language used,but have nothing to do with the length of the code.That is,a software has 20,000 lines of code and a software has 200,000 lines of code will has the same failure rate.Does anyone could explain it? Thank you.
Subject: Software Reliability
Reply Posted by: Bruce Dudley
Organization: Reliability Analysis Center
Date Posted: Wed Sep 22 11:09:10 US/Eastern 1999
In response to your questions on software reliability, the software expansion ratios delineate in the "Reliability Toolkit; Commercial Practices Edition" were developed prior to 1992 so would not include the languages such as VC++ and others listed. A worst case scenario for use with these languages would be to use the ADA language factor of 4.5 as a default condition. With regard to the other question, the size of the software program is not directly related to the reliability model presented, only the speed of the processor and number of instructions per unit time are significant. This model indicates the number of failures one should expect and the estimated failure rate after "x" program cycles and fault corrections. Software reliability is a difficult problem to estimate as one never knows how many faults exist at the start and how many more are introduced during correction of faults. Suggest that you review our "Introduction to Software Reliability A State of the Art Review" report for more information on software development and reliability. Another good source of information is the book "Ensuring Software Reliability" by Ann Marie Neufelder, Marcel Dekker, Inc., NY, NY, 1993.
Subject: Software Reliability
Reply Posted by: Birinder Singh
Date Posted: Thu Nov 11 11:33:48 US/Eastern 1999
the length of the program certainly does have impact on determing its failure rate. thesoftware that we are testing has shown that the estimation that we do based on the sample of failures detected can than be multiplied by two characterstics of the code to determine the remaining failures that are expected to be found. the two charaterctics are :
(i) lines of code
(ii) complexity factor