ARCHITECTURE OF A SELF-HOSTED BLOCKCHAIN FORENSICS SYSTEM FOR SOLANA: A PILOT EVALUATION OF A RUST IMPLEMENTATION

Abstract

Solana's high throughput and layered data model complicate the direct transfer of blockchain-forensics methods developed for Bitcoin and Ethereum. This study aims to design a self-hosted blockchain-forensics system implemented in Rust and provide a preliminary experimental assessment of its suitability for collecting, normalizing, and analysing Solana transactions. The research combines a thematic literature review with an assessment of a prototype based on read-only RPC access, explicit wallet-to-token-account owner resolution, a temporal multigraph, and explainable rules for wash trading, dusting, sandwich/MEV, and pump-and-dump patterns. The practical evaluation uses 10,000 mainnet-beta transactions from slots 250,000,000–250,001,000 on an AMD EPYC 7532 server with 64 GB RAM. Aggregate parsing times were 38 ms for Rust and 3,800 ms for Python, corresponding to calculated averages of 0.0038 and 0.38 ms per transaction and a 100-fold difference under the reported test conditions. Graph-construction times were 21 and 2,100 ms, while peak memory values were 42 and 480 MB, respectively. Because the implementations also differed in thread count, the result characterizes the compared software configurations as a whole and requires independent replication. The contribution is system-integrative: a testable blockchain-forensics pipeline for Solana that separates observed transaction facts, heuristic inferences, and an auditable evidence bundle.