The data set includes 72 hours of system measurement data as well as the response time performances for image classification systems.
The same image classification program is executed on either a cloud or an edge system with three different workload settings.
We conducted the same experiments four times (R1, R2, R3, and R4).
The obtained data are archived separately in the files ({R1, R2, R3, R4}-{Cloud, Edge}.zip).
We observed software aging phenomenon in these data sets. The details will be available in our paper currently under review.
The data sets are available for further research purposes.
E. Andrade, F. Machida, R. Pietrantuono, and D. Cotroneo, Software aging in image classification systems on cloud and edge, IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), pp. 342-348, 2020. [paper]
Last update: 2021.3.31