Reproducible dependability benchmarking experiments based on unambiguous benchmark setup descriptions

Benchmarking the Resilience of Self-Adaptive Systems: A New Research Challenge

“…DBench was a major project on dependability benchmarking. Several dependability benchmarks were defined in the context of the DBench project, targeting the following application domains: generalpurpose operating systems [12][13], real time kernels and control applications [17], on-line transaction processing systems [23] [5], and web-servers [8]. Other works on dependability benchmarking were carried on afterwards, but DBench did settle the foundations for many other initiatives (see [14] for an overview on dependability benchmarking initiatives).…”

Benchmarking the dependability of Windows NT4, 2000 and XP

“…The prototype we have developed uses a TPC-C client [14] as a privileged workload to be in conformance with the experiments performed on transactional systems in DBench [15,16]. We simply use the TPC-C client as a workload, but we do not use the performance measures specified by TPC-C as they are far from being suitable to characterize the behavior of an OS.…”

Hardware fault injection with UMLinux

“…In over 10 percent, the database failed completely, in a few cases database errors were logged. In extremely few cases the database faulttolerance mechanisms detected and correctly recovered from the error [1].…”