MODELING CORRELATION IN SOFTWARE RECOVERY BLOCKS

This paper considers the problem of accurately modeling the software f ault-tolerance technique based on recovery blocks. Models of such syst ems have been criticized for their assumptions of independence. Analys is of some systems have considered the correlation between software mo dules. This correlation may be due to a portion of the functional spec ification that is common to all software modules or due to the inheren t hardness of some problems. We consider three types of dependence whi ch can be captured using measurements. We consider correlation between software modules for a single input, correlation between successive a cceptance tests on correct module outputs and incorrect module outputs , and correlation between subsequent inputs. The technique we use is q uite general and can be applied to other types of correlation. In acco unting for dependence, we use the intensity distribution introduced by Eckhardt and Lee. We consider a new method of generating the intensit y distribution which is based on the pairwise correlation between modu les. This method provides us with a pessimistic result and a probabili ty-based approximation. We contrast this method with the assumption of independent modules as well as the use of the beta-binomial density w hich was introduced by Nicola and Goyal. For the purpose of obtaining numerical results, we use stochastic reward nets (SRN's) that incorpor ate all of the above dependencies and then use a modeling tool called Stochastic Petri Net Package (SPNP).