This paper presents an integrated software fault injection environment (DOCTOR) which is capable of (1) generating synthetic workloads under which system dependability is evaluated, (2) injecting various types of faults with different options, and (3) collecting performance and dependability data. A comprehensive graphical user interface is also provided. The software-implemented fault-injection tool supports three types of faults: memory faults, CPU faults, and communication faults. Each injected fault may be permanent, transient or intermittent. A fault-injection plan can be formulated probabilistically, or based on the past event history. The modular organization of tools is particularly designed for distributed architectures. DOCTOR is implemented on a distributed real-time system called HARTS , and its capability has been tested through numerous experiments.
|Number of pages||10|
|Publication status||Published - 1995 Jan 1|
|Event||Proceedings of the IEEE International Computer Performance and Dependability Symposium - Erlangen, Ger|
Duration: 1995 Apr 24 → 1995 Apr 26
|Other||Proceedings of the IEEE International Computer Performance and Dependability Symposium|
|Period||95/4/24 → 95/4/26|
All Science Journal Classification (ASJC) codes