Impact of Software Failures on the Reliability of a Man-Computer System

Abstract

Software failures and human errors are the most common reasons of inoparability of computer systems. Computers are increasingly reliable, but the level of transcient faults, caused by errors hidden in the programs, remains the same. For this reason software is becoming the key factor in the synthesis of highly reliable systems. Software errors do not result from operation, they either exist from the start or are inserted when patching or upgrading it. Some errors result from incorrect human interaction or unexpected environmental changes. Assessment of software impact requires functional-reliability approach to reliability analysis: the software fault occurs when it causes incorrect operation and not when it is introduced to the system. It is proposed to use software reliability models in system analysis, to predict the intensity of software faults. The software recovery after a failure is realized by restarts of various extend (microrestarts, minirestarts and macrorestarts). The impact of software failures and restarts on system availability is assessed.

Network Systems Analysis in Case of Critical Situations

“…Some changes may be the result of detecting system's faults and reacting to them. This is called system's reconfiguration [18,20]. The subsets of resources used by the tasks do not need to be disjoint.…”

“…The failures of the network structure -physical failures of technical infrastructure need to use adequate service teams, spare elements or substituted routes. Very often "technical" system's renewal processes are considered with assuming of the limited resources, for example the number service team for the part of the network [5,20].…”

“…The special services resources (service persons, different redundancy devices, etc.) supported by the so-called maintenance policies (procedures of the service resources using in purpose to minimise negative consequences of faults that are prepared before or created ad hoc by the system's manager) are build in every real system [3,4,17,20]. The maintenance policy is based on two main concepts: detection of unfriendly events and system's responses to them.…”

“…It is proposed to focus the dependability analysis of the networks on the fulfilment of requirements defined by user task [20]. Therefore, it should take into consideration following aspects:…”

“…If the management system could not create the service configuration then the service management process is started again and other task configuration may be appointed. These two decision processes are working in a loop, which is started up as a reaction on network events and incidents [3,4,5,20].…”