[PLACEHOLDER] Golang Gossipsub profiling and optimisation

[raulk]

Motivation and context

The current golang implementation of gossipsub is a reference implementation:

https://github.com/libp2p/go-libp2p-pubsub

It is intended to be correct and fully implement the protocol, but it is by no means optimized.

This leaves a lot of room for improvement in terms of performance; this more generally applies to the go-libp2p pubsub implementation.

We want to improve the reference implementation and make sure it is production-quality, so that it can be used in production environments without any fear or performance concerns.

Value

Benefits (if addressed)

ETH2 implementers can rely on a reference implementation to be used in production without any performance concerns.

Risks (if not addressed)

We are not addressing potential performance problems, which may cause issues in production.

Expected output and constraints

Relevant discussion

Design/implementation notes

Potential funders

[vyzo]

added a little bit of detail.

[daviddias]

@raulk @momack2 is this grant still relevant after the work done in https://blog.ipfs.io/gossipsubv1.1-eval-report-and-security-audit/ ?

[momack2]

Happy to deprecate this grant - agree I think the main aim was accomplished now 🎉