This project deals with addressing potential problems of outsourcing computations to cloud environments. We propose different approaches that are based on methods from dynamic anomaly detection to provide clients with trustworthy evidence about the runtime behavior of outsourced applications. In this setting, it is important that clients can check the evidence efficiently. In a hybrid cloud scenario, our approach permits (and in all generality requires) re-performing a small fraction of computations in the private cloud, the bulk of the workload must still execute in the public cloud for the approach to pay. Our approach can be fully automated, it requires no manual intervention.
We implemented our approach to BCC using the jp2-profiler and jborat tools. Logging the profiles is done using jp2-profiler.
We used modified versions of both tools jp2profiler and jborat to evaluate our approach with the pdfbox, POI-HWPF and POI-HSLF programs. Our code for compliance check and comparing the different approaches can be found here.
The test data can be obtained upon request.