Date: March 6, 2026
Title: FicusDB: Scalable Multi-Versioned Authenticated Archival Storage
Speaker: Hongbo Zhang
Abstract: As consensus protocols scale, storage has become the dominant bottleneck in modern blockchains. Systems must maintain historical versions, generate integrity proofs, and sustain high throughput as the state grows. Prior work often redesigns the authenticated data structure (ADS), but such changes sacrifice compatibility and require disruptive hard forks. We take the complementary approach: redesigning the storage layer while preserving the existing ADS interface. We present FicusDB, a log-structured archival storage system for copy-on-write tries. FicusDB introduces three innovations: (i) append-only logs with location-based identifiers that eliminate compaction, (ii) a CoW-aware tree-structured cache that avoids LRU thrashing, and (iii) the Aggregated Hash Array (AHA), which decouples proof correctness from storage durability by treating child-hash arrays as verifiable hints. Together, these techniques cut write amplification, improve read locality, and accelerate proof computation. On 11.5 million Ethereum blocks (≈1B keys), FicusDB achieves 3.7× higher storage throughput and a 66% smaller footprint compared to Geth, while remaining fully compatible. More broadly, FicusDB shows that substantial performance gains are possible without changing ADS semantics, reframing persistence itself as a locus of innovation for blockchain scalability.
Bio: Hongbo Zhang is a Ph.D. student in Computer Science at Cornell University, advised by Robbert van Renesse. He is interested in Computer Systems, with a current focus on building high-performance storage engines designed to handle the intensive versioning and authentication requirements of modern blockchain systems.