Gossip Observer: New project to monitor the Lightning P2P network

In the discussion on optimizing message propagation among peers in a network, a strategy is suggested that involves sending updates to peers at regular intervals, specifically every 12 seconds. This approach aims to keep the differences between each set of updates small, thereby reducing the amount of data that needs to be encoded and decoded, which in turn speeds up the process. The idea is to have the updates from one peer integrated before moving on to the next, although this might not always be feasible with a very large number of peers. However, even with this limitation, it is seen as a significant improvement over the traditional method of flood propagation.

The possibility of adding an additional communication round to address missing messages is considered, but it is noted that this could lead to inefficiencies by using up sketch capacity without adding value. A solution proposed involves directly incorporating the gossip message into the sketch element, avoiding the need for hashing, and thus making it easier to update the sketch with ongoing messages, especially when these are identifiable through specific criteria such as SCID and block height.

Furthermore, the discussion touches upon various trade-offs inherent in designing such a system, highlighting the implicit assumptions regarding the number of gossip peers, and the balance between computing power for sketch encoding and available bandwidth. To ground these considerations, it suggests setting clear design goals and understanding hardware limitations. For illustrative purposes, it proposes a scenario where the system operates with a number of channels and updates comparable to current standards, and the encoding/decoding tasks can be managed by a device with modest computational capabilities, like a Raspberry Pi 5, assuming ten peers with one round of communication per minute as a baseline. This example serves to contextualize the theoretical discussion with a practical benchmark.