Blockchain Nodes are Heterogeneous and Your P2P Overlay Should be Too: PODS

TL;DR

This paper introduces PODS, a flexible P2P overlay architecture for blockchain systems that accommodates node heterogeneity through multiple domains, improving scalability and security.

Contribution

It proposes PODS, a novel multi-domain P2P overlay model, and designs two new node discovery protocols, FedKad and SovKad, demonstrating their advantages over existing methods.

Findings

SovKad outperforms single and multi-overlay architectures in hop count and success rate.

FedKad uses slightly less bandwidth than SovKad.

PODS enhances blockchain P2P networks by supporting heterogeneous node groups.

Abstract

At the core of each blockchain system, parties communicate through a peer-to-peer (P2P) overlay. Unfortunately, recent evidence suggests these P2P overlays represent a significant bottleneck for transaction throughput and scalability. Furthermore, they enable a number of attacks. We argue that these performance and security problems arise because current P2P overlays cannot fully capture the complexity of a blockchain system as they do not offer flexibility to accommodate node heterogeneity. We propose a novel approach to address these issues: P2P Overlay Domains with Sovereignty (PODS), which allows nodes in a single overlay to belong to multiple heterogeneous groups, called domains. Each domain features its own set of protocols, tailored to the characteristics and needs of its nodes. To demonstrate the effectiveness of PODS, we design and implement two novel node discovery protocols: FedKad and SovKad. Using a custom simulator, we show that node discovery using PODS (SovKad) architecture outperforms both single overlay (Kademlia) and multi-overlay (FedKad) architectures in terms of hop count and success rate, though FedKad requires slightly less bandwidth.