CISUC

Plug and Play Fault Injector for Dependability Benchmarking

Authors

Abstract

The increased reliance of our society on computers requires that their dependability be measured. Benchmarks with that objective must include fault injectors that are very simple to use. They should be downloadable from the web together with the other components of the benchmark, be able to target all system components, and should not require any complex setup or installation procedure. Since the existing injectors did not fulfill this requirement, we have developed DBench-FI, a fault injector for dependability benchmarking that is very simple to install and use. DBench-FI belongs to the software implemented fault injection (SWIFI) family, since no special hardware can be used for this purpose. The current version targets the Linux OS on Intel processors, and uses a flexible runtime kernel upgrading algorithm to allow access to the target process memory space, that can be in either user or system space, enabling in this way the injection of faults. Furthermore, it does not require the availability of the source code of any system component or user process. It is not based on any debug mode either, being able to inject faults even in tasks that were already running when it is installed, irrespective of their complexity, like the Oracle Database Management System that is used in the examples in the paper. This set of features is, to the best of our knowledge, unique. It currently only injects memory faults, but there are plans to include other fault models in the future. In order to demonstrate the capability of the DBench-FI tool, the results of complex fault injection campaigns are presented.

Keywords

SWIFI, Dependability Benchmarking

Subject

Software Fault Emulation

Related Project

DBench - Dependability Benchmarking

Conference

First Latin-American Symposium on Dependable Computing, LADC, October 2003


Cited by

Year 2013 : 1 citations

 Stefan Winter, Michael Tretter, Benjamin Sattler, and Neeraj Suri, “simFI: From Single to Simultaneous Software Fault Injections”, The 43rd Annual IEEE/IFIP International Conference on Dependable Systems and Networks, Budapest, Hungary, June 2013.

Year 2011 : 1 citations

 1. Gianpiero Cabodi, Marco Murciano, Massimo Violante, “Boosting software fault injection for dependability analysis of real-time embedded applications”, ACM Transactions on Embeded Computing Systems, Vol. 10, Issue 2, January 2011.