Gossip Observer: New project to monitor the Lightning P2P network

In the realm of peer-to-peer (P2P) networks, particularly those related to payment systems, a significant observation has been made regarding the propagation of messages across nodes. Despite expectations for high connectivity due to the default number of connections implementations make, which is five or more, there are instances where nodes miss messages pertinent to entire subgraphs or neighborhoods within the payment network. This phenomenon raises questions about the possibility of the gossip graph being partitioned and whether the implementation of three random gossip syncers could ensure a unified network, assuming all nodes operate reliably.

One critical point of discussion revolves around bandwidth usage alongside other factors such as reducing message delays. Measurements indicate that approximately 95% of nodes receive a given message after 600 seconds, although this only accounts for nodes that forward the gossip message to the observer node. With an observer node reaching 900 concurrent connections at peak times but about 1,250 nodes with known IP socket addresses in the channel graph, the inaccessibility of 28% of IP nodes becomes a concern. Misconfiguration of routers or firewalls, especially among nodes announcing both Tor and clearnet addresses, may hinder inbound IPv4 connections despite broadcasting an IPv4 address. Efforts from Bitcoin Core to address these issues through NAT-PMP / PCP support, following the deprecation of UPnP support, highlight the challenges in ensuring node reachability and liveness.

Further exploration into node liveness through a different study revealed that out of 16,626 peers in the channel graph, an average of 4,195 peers were connected over a 10-day period, with 4,859 nodes generating at least one new gossip message during this time. Over a broader timeframe of 55 days, 7,579 nodes were reachable at least once, despite attempts to connect to 13,165 nodes. The presence of 5,607 private nodes in the channel graph, which do not advertise socket addresses, complicates the reachability issue. The discovery of nodes not frequently involved in network communication raises doubts about their ability to process payments and contributes to the debate on optimizing bandwidth usage and redundancy in message updates.

The analysis of message propagation efficiency within the network reveals disparities in how messages are received by nodes. A study from the gossip_observer repository identified varied peaks in message reception, with most messages failing to reach more than 500 of the nodes. This inconsistency suggests different propagation behaviors for specific message types and emphasizes the importance of network reliability over convergence delay. The investigation into staggered broadcast mechanisms and the inclusion of deduplication methodologies offers insights into improving message dissemination across the network.

Addressing these concerns requires a nuanced understanding of P2P network dynamics, focusing on enhancing the reliability of message sending and ensuring comprehensive network coverage. The conversation underscores the complexity of maintaining efficient and inclusive communication channels within decentralized networks, pointing towards ongoing research and adjustments in network strategies to foster improved connectivity and functionality.