Items to note
- How we imagine economy working between Saturn and SPs
- Saturn - logging each retrieval and payments
- Economic changes won’t be a core protocol change
- Good case for paid retrieval, but maybe farther out
- how do you incentive off chain transactions ???
- think this thru - integrating retrieval networks with SP network
- timeline
- use case
- selector queries will be useful
- will immediately hit problems if selectors are not there
- how often will things be cache missed? → every file will be cache missed at least once unless they do a smart cache warming
- trying to match IPFS load around CIDs requested and also traffic
- trying to figure out smart cache missing to other retrieval markets etc. rather then going to SPs - will be difficult to enforce this for other retrieval networks
- ttfb on cache miss requests (still need to serve to end client)
- order of seconds (ish) ideally better
- 11.7 seconds is current IPFS gateway performance (95th percentile)
- longer term: Filecoin content accelerated
- need to tighten up longer term product
- assume for the forseeable future that it will be FIL+ data
Retrieval Networks
- also partner wanting to cache miss to SPs (Titan, Myel, etc.)
Potential next steps