Transaction Conflict Control in Hyperledger Fabric: A Taxonomy, Gaps, and Design for Conflict Prevention

The execute-order-validate approach to blockchain consensus, most notably implemented by Hyperledger Fabric, facilitates highly scalable execution of smart contract - in Fabric terminology, "chaincode" - invocations in cross-organizational blockchains; at the expense of requiring multi-version concurrency control conflict handling during block validation. Consequently, the system-level goodput can be significantly lower than throughput. Although several solutions have been proposed for handling and avoiding conflicts in Hyperledger Fabric, a systematic and holistic approach is missing. We introduce the notion of conflict-controlled operation, propose a novel taxonomy of its means based on the codified principles of dependable computing, and categorize the known approaches. Based on this taxonomy, we identified the critical gaps in the state-of-the-art. Design-time conflict prevention is one such gap, and we propose the application of a model-driven engineering process for this purpose. For the last storage mapping stage of the process, we propose entity attribute partitioning for conflict prevention, describe a data mapper-style chaincode layer, and empirically evaluate our solution.