On the Storage Overhead of Proof-of-Work Blockchains

TL;DR

This paper investigates the storage overhead of Proof-of-Work blockchains, demonstrating that full nodes can significantly reduce their storage requirements to around 15 GB through empirical analysis and client-side strategies, without protocol changes.

Contribution

It provides a thorough empirical analysis of blockchain data usage and proposes practical methods for reducing storage overhead without altering the protocol.

Findings

Full nodes can store approximately 15 GB instead of hundreds of GB.

Empirical analysis reveals how nodes utilize ledger data for validation.

Client-side strategies can further minimize storage with minimal computational cost.

Abstract

Permissionless blockchains such as Bitcoin have long been criticized for their high computational and storage overhead. Unfortunately, while a number of proposals address the energy consumption of existing Proof-of-Work deployments, little attention has been given so far to remedy the storage overhead incurred by those blockchains. In fact, it seems widely acceptable that full nodes supporting the blockchains have to volunteer hundreds of GBs of their storage, to store and verify all transactions exchanged in the system. In this paper, we explore the solution space to effectively reduce the storage footprint of Proof-of-Work based blockchains. To do so, we analyze, by means of thorough empirical measurements, how existing full blockchain nodes utilize data from the shared ledger to validate incoming transactions/blocks. Based on this analysis, we show that it is possible for full nodes to locally reduce their storage footprint to approximately 15 GB, without any modification to the underlying protocol. We also discuss other client-side strategies to further reduce the storage footprint while incurring negligible computational overhead on the nodes.