• Proposed by @Adin Schmahmann
• Completed as part of: Provider Record Liveness
• Final Report: https://github.com/protocol/network-measurements/blob/master/results/rfm17-provider-record-liveness.md

The IPFS Public DHT has maximum lifetimes for certain records (provider records 24hrs, IPNS records 36hrs). Record lifetimes serve a dual purpose a dealing with network churn and having some work symmetry between clients and servers for requests. However, it would be useful to understand how long data could be discoverable in the network if record lifetimes were infinite.

We could determine this fabricating some identifiers in network (XOR) space and seeing over time how many of the 20 closest peers at the beginning of time are still within the 20 closest peers over time.

Note: this isn’t quite perfect because when we do Provides/Puts we are not currently sending to absolutely the 20 closest peers. However, this can give us some initial intuition to work off of that in the future we can augment after seeing what percentage of the absolutely closest 20 peers our client implementations actually send messages to.

Tools to help here:

• Nebula Crawler Database
• https://github.com/libp2p/go-libp2p-xor (the Trie in particular)