Is it correct to say that the most famous 25 relays are being basically DDoSed all the time while the other 900 relays are seeing very little traffic?

Reply to this note

Please Login to reply.

Discussion

Yes

The best two to answer this that I know of atm are #[2] and #[3]

Thanks for bringing this up, here is a chart that's long overdue. Events/pubkeys per relay in the last 7 days.

https://stats.nostr.band/#relay_writers

I can't measure their read-load obviously, but can tell how much events are written. As I expected, the list of top relays by number of events is materially different from list of top relays by contact list. Does this tell you anything new/interesting?

Whoa, I had not seen these stats, they are amazing.

I built them this morning after seeing this thread

Cool! How do you define a “high-quality” pubkey?

There is a ref and explanation at the top of the page. Pubkeys with non zero trust rank

oh did not see that, thanks!

Probably not far off, yeah. That’s why it would be beneficial to continue to make both relays and clients smarter until relays can be used as interchangeable dumb pipes, without taking a hit on coverage of content.

One way (which I know not everyone agrees with me on) is to have the Nostr equivalent of Mastodon's "instance federation" system.

In other words, for all relays who chose to do so, they would sync with every other known relay using relay software that supports that, like strfry.

This way, no matter which relay a user connects to, as long as that relay uses this functionality, they'll have good coverage of the posts of the users they follow.

If most relays do this, then it'll be possible for clients to intelligently pick which relays they connect to based on evenly distributing the load across relays, and they won't have to worry about missing out on events.

Zipfs law in play I guess

power law

New users generating keys on snort get random relays, so I hope not

how many relays in the pool?

List is pulled from nostr.watch online relays

nice 👌

Given the current set of incentives in the protocol, it seems like this is the only possible outcome. Anything that makes relays not interchangeable has to end up this way, right?

If relays were like Internet routing hardware where nobody even knew which relay they were connected to (who knows which brand of BGP router their traffic flows through?) then using some kind of refined version of the gossip protocol could (maybe?) be able to spread the load.

Finally, had to add wss://relay.snort.social/ with the slash, somehow it shows separate under relays without the slash, to see this message, and why don’t you have relays shown on your profile, is this some privacy feature?

😃😃😃

okay https://relay.nostrdocs.com (paid) is here to take some of the heat. Hit it

How have other P2P networks with diverse clients solved this? Torrents, I2P, IPFS...

have compatibility for old and new

Nostr isn’t P2P

I meant it in the decentralized sense. BitTorrent has trackers similar to relays. The issue of getting good tracker for a torrent is similar to getting good relays on Nostr.

Current nostr client ecosystem is not encouraging diverse relay selection which could lead to some relays becoming so large that the service becomes centralized.

Of course that will happen. Same thing happened to the Fediverse. Just wait for the relay blocking

Exactly I don't know what we have waiting for.

I just started using NOSTR and my first experience is with astral.ninja. At first sight, I've got lots of SPAM Chinese stuff on the global feed. What is going on?

Can speak from BRB, non stop pounding. But does need some upgrades that could resolve most of the performance issues.

we could have a pubkey:relay dht and maybe add pubkeys to note links

So things seem to be going according to my predictions.

I expect it to get worse as user base further grows.