Stu on Nostr: As more of us follow more and more people the server load grows quite steeply ...
As more of us follow more and more people the server load grows quite steeply doesn’t it?
Every time someone refreshes, each relay queried has to look up each pubkey on that persons follow list.
If you follow lots of people, your refreshes demand more relay load to find the notes.
10,000 users in the TestFlight means there is a ceiling of 100m follows in the user base. We will never get close to that, but even so, as the current user base gets more interconnected the relay load will grow, even as the number of users stays flat.
It’s not the number of users that define the relay load, it’s the number of follows all the active users have.
Should there be a ceiling on how many people you can follow? Otherwise it could be a potential vulnerability.
I guess one model for monetising a relay, is to demand a user follow that relays pubkey in exchange for using the relay. I can see my followers so I assume a relay knows who followers it’s pubkey. It can then whitelist serving requests from the pubkeys that follow the relay pubkey.
This then gives the relay an effective broadcast reach that it can sell (ads. Sorry! 😭) to pay for the cost of hosting a beefy and scalable relay (eg Lambda).
The other option is to pay a sub to a fast premium relay.
Alternatively some clients (apps), might choose to run their own relays which are coded into their clients by default.
I can imagine several models to monetise relays. None of them are too complicated and I think scale might soon put a huge amount of demand on popular relays which will require substantial beef (and hosting costs) to handle the load.
Every time someone refreshes, each relay queried has to look up each pubkey on that persons follow list.
If you follow lots of people, your refreshes demand more relay load to find the notes.
10,000 users in the TestFlight means there is a ceiling of 100m follows in the user base. We will never get close to that, but even so, as the current user base gets more interconnected the relay load will grow, even as the number of users stays flat.
It’s not the number of users that define the relay load, it’s the number of follows all the active users have.
Should there be a ceiling on how many people you can follow? Otherwise it could be a potential vulnerability.
I guess one model for monetising a relay, is to demand a user follow that relays pubkey in exchange for using the relay. I can see my followers so I assume a relay knows who followers it’s pubkey. It can then whitelist serving requests from the pubkeys that follow the relay pubkey.
This then gives the relay an effective broadcast reach that it can sell (ads. Sorry! 😭) to pay for the cost of hosting a beefy and scalable relay (eg Lambda).
The other option is to pay a sub to a fast premium relay.
Alternatively some clients (apps), might choose to run their own relays which are coded into their clients by default.
I can imagine several models to monetise relays. None of them are too complicated and I think scale might soon put a huge amount of demand on popular relays which will require substantial beef (and hosting costs) to handle the load.