jeff on Nostr: What if relays split into three types? indexing relays - Clients post and query ...
What if relays split into three types?
indexing relays - Clients post and query profiles and relay information. These would be much more economically cheaper to run, search, etc. Node cost is inversely proportional to node. You could run this over POST and GET, and results could be turned into static files, which can be served wicked fast. People would be expected to re-index this, and propagate it, across the globe. UX would speed up huge, cause regular relays wouldn't be getting hit with "Hey, do you have data about this user, no? oh - okay". API could be far richer too.
read-relays - Optimized for search, filtering, serving. Anybody can still write, but the writes would be slow / more-expensive.
write-relays - Optimized for publishing, and fanning out to the read-relays. Anybody can still read, but the reads would be slow / more-expensive.
- Jeff
indexing relays - Clients post and query profiles and relay information. These would be much more economically cheaper to run, search, etc. Node cost is inversely proportional to node. You could run this over POST and GET, and results could be turned into static files, which can be served wicked fast. People would be expected to re-index this, and propagate it, across the globe. UX would speed up huge, cause regular relays wouldn't be getting hit with "Hey, do you have data about this user, no? oh - okay". API could be far richer too.
read-relays - Optimized for search, filtering, serving. Anybody can still write, but the writes would be slow / more-expensive.
write-relays - Optimized for publishing, and fanning out to the read-relays. Anybody can still read, but the reads would be slow / more-expensive.
- Jeff