EIP8025 libp2p specification

[frisitano]

Overview

There has been discussion around how ExecutionProofs should be gossiped on the p2p network. The high-level conclusion reached is that nodes should announce an IHAVE message for an ExecutionProof that they can serve. It is then the responsibility of their peers to request the proof. The intention is to minimise redundant bandwidth consumption by preventing eager gossiping of execution proofs to peers. This is particularly important because execution proofs are large (currently 300-400KB).

We should determine how best to capture this in the consensus specs.

[frisitano]

For additional context, here is a recap of prior discussions on proof gossip networking (announce / IHAVE / fetch trade-offs, bandwidth considerations, peer selection):

https://hackmd.io/@frisitano/H1XJS3XTZx

The note pulls together the threads referenced in this issue's overview and aims to provide a basis for converging on an approach.

[pheobeayo]

@frisitano Can I take this issue up?

[frisitano]

@pheobeayo Yes, please feel free to work on this. If you could suggest a plan on how we can best capture this in the specs, that would be great. We can then proceed with the actual implementation.

[pheobeayo]

@frisitano I am on it.