Otimização do desempenho de hashes com parallelhash para aceleração do hyperledger fabric

Abstract

Blockchain technology ensures data integrity and traceability in distributed environments, relying on cryptographic hash functions to safeguard blocks and state data. In permissioned frameworks like Hyperledger Fabric, this reliance creates performance bottlenecks as data volume increases, since conventional algorithms (e.g., SHA-256) execute sequentially and underutilize multicore architectures. This work investigates ParallelHash (NIST SP 800-185) to mitigate this bottleneck in the Fabric snapshot process. A Go implementation using goroutines was developed and evaluated in two scenarios: isolated benchmarks and tests with real snapshot artifacts. Results indicate a 47% reduction in total time and a 1.9x throughput increase in the hashing stage for medium and large inputs. The study also proposes a modular integration strategy compatible with the existing architecture. Conclusions highlight the technical viability of ParallelHash in intensive validation scenarios, adhering to cryptographic agility principles.