Drive IPFS file-transfer speedups

[adlrocha]

We are working on collecting and organizing all the information around file-sharing in P2P networks and the current implementations of content exchange protocols in IPFS (mainly Bitwswap and Graphsync) so that we have a foundation for the design, proposal and implementation of further improvements over file-sharing in IPFS.

All our contributions are collected in this document (I thought it was a better format than a long Github issue). It is by no means complete or in a final state, and we hope to keep it "alive" to include all the feedback we can receive from the community.

In this sense, we would love to gather your insights and ideas. If you can come up with important use cases we may be missing,
proposal ideas from the community we may have not come across, common requesting patterns we may not have considered, or complement something, please feel free to comment here or in the document.

This work is intended to be an extension to the Bitswap/Graphsync Open Problem, and the foundation for future improvements built on top of this amazing work.

[adlrocha]

All the efforts being around driving speed-ups for file-sharing in IPFS and beyond have been aggregated under the umbrella of what we've called the "Beyond Bitswap" (or Beyond Swapping Bits) project. We've summarized all the contributions so far in this post, and all the assets and documentation of the project can be found in this repo.

The repo includes a list of RFCs at different stages with ideas and prototypes on how to improve file-sharing. There are already discussion going on in the issues and PRs with new ideas and proposals, and further RFCs are being prototyped.

The repo also includes a testbed to easily benchmark and debug file-sharing protocols over IPFS. There is a lot of work and good ideas over there, so do not hesitate to give it a look!