Oddbean new post about | logout
 It is possible, but the main problem is the underlying data substrate. You need something that scales well and horizontally and does not rely on a health of a single server. The problem, nobody besides one person that I know of, is working on it. It should also be economically viable, so not just using some expensive DBaaS 🐶🐾🫡 
 I would run a few relays if I could just spin up some droplets in DO 
 More relays is not the solution, client has finite resources and expensive bandwidth. 🐶🐾🫡 
 Please expand on this if you have time 
 Relay is just a DB, that’s all. You have a DB that has a JSON WS interface that client connects to. Having more DBs just increases the number of small DBs that run in the vacuum. If nobody writes to it then it is useless. 🐶🐾🫡 
 you can install a relay hub on any linux machine, docs here:  https://github.com/relaytools/docs