The relays have faced great challenges in recent months with the arrival of waves of users, the last, and most evident, China and Hong Kong.

The relays are computers, they require memory, CPU and RAM. Although the essence of a relay is that it be "dumb", to face the scalability challenges we will need a much larger network.

The Clients have not behaved well with the relays, they have demanded a lot of information from the relays to improve the user experience. Some request the last 400 notes so that the user can scroll without affecting their experience and this is fine for a small user network, but with the different waves of users that have arrived in Nostr the relays have faced the need to scale infrastructure, assuming significant costs, to meet the demand of new users.

Although the Clients should optimize their request for information to the relays, the owners of these computers must still bear the costs of supplying the user demand.

How can we scale in infrastructure without costs being absorbed for free by one person?

The relays have come to privatize their use, we have seen in recent weeks the emergence of at least 10 private relays, whose cost is between 1,000 sats and 15,000 sats, but even this would not be able to supply the cost of maintaining a private relay with costs that are around the 160 USD per month.

I believe that there is a transitory and scalable solution to the problem of the maintenance cost of the private relay and it is the construction of Relay federations where the Relay dominator takes the resources of relays distributed in the globe contributing computational and storage resources and thus distribute the cost of maintenance.

In order for the popular Relays to implement their federation, a simple solution must be brought to them so that collaborators can easily and quickly connect their hardware to this dominant Relay.

This would be one of many ways that we can implement to prepare for the next wave of users that is coming.

There is a limit to how long Relay Owners can last.