Diagnostics in Software-defined Infrastructure on the SAVI testbed

With the increasing complexity of data center networks, the operations, management and diagnosis of data centers is a critical problem. The need of a self-managing, self-configuring, and self-optimizing network is crucial which is known as Autonomic Networking concept. In this thesis, we propose an autonomic Diagnostics framework in Software-Defined Infrastructure running in the SAVI testbed. We have analyzed and discussed various design requirements and open challenges in literature. We have leveraged Graph Mining and Machine Learning techniques in our approach in order to detect different kinds of anomalies. We have studied different algorithm performance under different settings for detecting anomalies. We have tested our framework on several use cases: Standalone Webserver, Standalone Database, Webserver-Database workload pattern, bandwidth throttling between a pair of VMs, denial-of-service attack on a Webserver and Spark Job failure. Our framework was able to detect the aforementioned anomalies accurately.M.A.S