Workload Characterization and Performance Evaluation of a Blockchain Implementation for Managing Federated Cloud Resources - Assuming a Peer-to-peer Energy Management Use Case

Abstract: Blockchain technology has become an appealing concept in Distributed Systems because it enables a distributed storage of information, replacing a central database [1]. In addition, Blockchains promise to address inherent and difficult issues in distributed systems such as a) proving the provenance of information, i.e., the documentation where pieces of data comes from (including their the processing), and b) that the information has not been changed, i.e., the integrity of the information has not been corrupted. The data in a Blockchain is said to be immutable. In this thesis, we apply Blockchain technology as a concept in Distributed Systems for securely collecting and storing data from distributed cloud resources that must be intact over a longer amount of time, such as the amount of consumed cloud resources characterized by CPU load or energy usage. In particular, this work considers a peer-to-peer energy use case where virtual energy resources are monitored. The focus of this thesis is on a) how a Blockchain for a distributed Cloud monitoring can be implemented, b) how the workload can be characterized and c) how the Blockchain system’s performance can be observed and what performance can be achieved. Therefore, the work defines an initial system model, provide an implementation, and carries out experiments in order to understand the impact of the design factors and the system input to the capabilities and performance of the system. The results of the experiments, the workload characterization and performance analysis, are analysed by statistical means and provided as graphs. The choices of system models, Blockchain technology (Hyperledger Fabric), and other parameters, are based on the literature review. The experimental implementation is, in turn, based on the selected system model, where we want to experimentally identify limitations and bottlenecks of the performance.