| T. Anderson, P. A. Barrett, D. N. Halliwell, and M. R. Moulding. An evaluation of software fault tolerance in a practical system. In Fault Tolerant Computing Symposium 15, pages 140--145, Ann Arbor, MI, June 1985. IEEE Computer Society. |
....effort, cost of delays and so on. The cost of diversity has been discussed e.g. in [Migneault 1982] Laprie et al. 1990] Voges 1994] Some information on this cost effectiveness issue albeit incomplete comes from experiments that have been conducted under controlled conditions. In [Anderson et al. 1985] it is reported that a 2 version recovery block system masked about 70 of the failures that would have taken place in the single channel of the primary version. That is, the 2 version system is less than half as unreliable as the single version. We might be tempted to assume that a 2 version ....
T. Anderson, P. A. Barrett, D. N. Halliwell and M. R. Moulding, "An Evaluation of Software Fault Tolerance in a Practical System", in 15th IEEE Int. Symp. on Fault-Tolerant Computing (FTCS15) , Ann Arbor, Mich., pp.140-145, 1985.
....is used to detect errors and locate failed modules. An alternative to diagnosing faults by voting is to use a single component and monitor its output. This is the basis for a software technique called recovery blocks which detect software faults through the use of an explicit acceptance test [Anderson85, Horning74] First proposed by Horning, Lauer, Melliar Smith, and Randell, recovery blocks do not concurrently execute multiple software modules but instead monitor the output of a single primary module. If the acceptance test detects an error, a secondary module is called upon to replace the ....
Anderson, T., Barret, P. A., Halliwell, D. N., and Moulding, M. R. "An evaluation of software fault tolerance in a practical system." Proceedings of the 15th Annual International Symposium on Fault-Tolerant Computing (FTCS-15). Los Alamitos, CA: IEEE Computer Society Press, Ann Arbor, MI (June 19-21, 1985) 140-145.
....(or database transactions) are partially correct. Clearly, the use of automatic backward recovery improves the chance that crucial data will remain consistent in the presence of failure detections, since it provides tolerance for all confined design faults. Experimental studies confirm this [Anderson85] However, to acquire confidence that a recovery block is capable of tolerating all design faults that might be contained in its alternates and acceptance test is in fact as hard as proving that these alternates together with the acceptance test are partially correct. DISCUSSION This paper ....
....program versions for a single specification so that the failure domain of the resulting multi version program is smaller than the failure domains of the individual program versions used. Empirical investigations of the likelihood of this goal being achieved for actual programs can be found in [Anderson85] Eckhardt91] The second part of the paper investigates what is exception handling in programs structured as hierarchies of data abstractions. The answer proposed is a simple one. At each level of abstraction, exception handling consists of: detection, attempt at masking, consistent state ....
T. Anderson, P. A. Barrett, D. N. Hallivell, M. R. Moulding, "An evaluation of Software Fault Tolerance in a Practical System", in Proceedings 15th International Symposium on Fault-Tolerant Computing, pp. 140-145, Ann Arbor, Michigan, 1985.
No context found.
T. Anderson, P. A. Barrett, D. N. Halliwell, and M. R. Moulding. An evaluation of software fault tolerance in a practical system. In Fault Tolerant Computing Symposium 15, pages 140--145, Ann Arbor, MI, June 1985. IEEE Computer Society.
No context found.
T. Anderson, et al, An Evaluation of Software Fault Tolerance in a Practical System, Digest of Papers FTCS-15: The Fifteenth Annual International Symposium on Fault-Tolerant Computing, June 1985, pp. 140 - 145.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC