Stateful to Stateless: Modelling Stateless Ethereum

Sandra Johnson, David Hyland-Wood, Anders L. Madsen, Kerrie L. Mengersen

Abstract

As Ethereum’s state grows indefinitely, it threatens the network's decentralization by increasing the resource requirements for full nodes. "Stateless Ethereum" aims to address this by introducing witnesses—Merkle proofs that allow nodes to verify blocks without storing the full state. This paper presents an Object-Oriented Bayesian Network (OOBN) to model the transition to a stateless architecture. Our findings indicate that while witnesses introduce significant data overhead, the Ethereum ecosystem is likely to remain healthy, with a 56% probability of maintaining a functional state under the proposed changes.

Introduction

Ethereum currently operates in a "stateful" manner, requiring every full node to store the entire history and current state of the network, including account balances and smart contract code. This state grows "ad infinitum," leading to high storage costs and long synchronization times, which creates a barrier to entry and risks centralizing the network.

"Stateless Ethereum" is a proposed solution that introduces witnesses—data packets containing Merkle proofs that allow a node to verify a block without needing the full state locally. However, witnesses add significant data overhead to every block. This research addresses a critical question: Will the increased network traffic from these witnesses degrade the health and efficiency of the Ethereum ecosystem?

Methodology

The authors employ Object-Oriented Bayesian Networks (OOBNs) to handle the uncertainty and complexity of the blockchain ecosystem. This probabilistic graphical modeling approach allows the system to be broken down into four modular, hierarchical sub-models: Block Creation, Witness Creation, Ethereum Network, and Block Propagation.

The high-level OOBN structure illustrating the modular relationship between block creation, witness generation, and network propagation.

The model was quantified using a hybrid approach: empirical data from the Ethereum Mainnet and network statistics were used where available, while expert elicitation was employed to fill data gaps, particularly regarding the impact of node location on peer-to-peer latency.

Datasets

  • Ethereum Mainnet Data: 26,595 blocks mined in late 2019, used to learn the structure and conditional probabilities of block and witness creation.
  • Network Statistics: Data from ethernodes.org (node types and distribution) and Wondernetwork (global latency data).
  • Propagation Data: Historical block propagation times sourced from ethstats.io and academic studies.

Witness Creation and Network Impact

A core component of the transition is the generation of witnesses. The model quantifies the size and time required for witness creation based on block difficulty and state entries updated.

The Witness Creation sub-model, detailing the probabilistic dependencies between block parameters and the resulting witness size and creation time.

By comparing the model with and without witnesses, the researchers found a relative 7% decrease in "health" (from 60% to 56%) when witnesses are introduced. This quantification highlights the trade-off between lower local storage requirements and higher global bandwidth usage.

Key Insights and Results

  • Ecosystem Health: The Ethereum ecosystem is expected to remain healthy following the introduction of Stateless Ethereum, with the "healthy" state being the most probable outcome (56%).
  • Sensitivity to Node Status: The model revealed that "Node Status" (whether a node is up-to-date, syncing, or offline) and the ability to "Keep up with the head of the chain" are the most influential factors on overall network health.
  • Worst-Case Scenarios: In scenarios involving very large witnesses and non-mining nodes, the probability of a node keeping up with the chain dropped from 65% to 54%—a 17% relative decline.
  • Propagation Bottlenecks: The model confirmed that network latency and node bandwidth remain the primary constraints on how quickly the network can reach consensus in a stateless environment.

Conclusion and Future Directions

The study acknowledges several limitations, including the use of data that predates EIP-1559 and the omission of newer technologies like Verkle tries, which would significantly reduce witness sizes. Future work should incorporate these technological evolutions and explore the interaction between statelessness and other proposals like "Regenesis" (periodic state pruning).

Despite these limitations, the OOBN framework provides a robust, modular tool for researchers to evaluate the impact of protocol changes on the Ethereum ecosystem's long-term health.

Summary updated

BibTeX
@inproceedings{johnson2022stateful,
  author = {Sandra Johnson and David Hyland-Wood and Anders L. Madsen and Kerrie L. Mengersen},
  title = {{Stateful to Stateless: Modelling Stateless Ethereum}},
  booktitle = {MARS@ETAPS},
  year = {2022},
  url = {https://openreview.net/forum?id=Ndm0diDn3AX},
}
Figure