500M tweets/day (and growing) is 5787 tweets/sec. What is likely TPS for final tweet-txns (typos4evar ;-) of Solana given the contention from worst-case and average user load? Not 50KTPS.
This Tweet was deleted by the Tweet author. Learn more
Conversation
Quote Tweet
Replying to @BrendanEich @heikkiarponen and 3 others
No such blockchain exists. The scale is daunting: omnicoreagency.com/twitter-statis: 500M tweets/day. This begets centralization and advertising as implicit revenue model. People generally won’t pay in an explicit quid pro quo arrangement; at least no one has demonstrated many will do so.
9
3
39
This Tweet was deleted by the Tweet author. Learn more
This Tweet was deleted by the Tweet author. Learn more
yeah really the answer here is we should build it and then spam a ton of tweets and see what happens :P
1
3
How about some final TPS under load numbers? We don’t want impostors spoofing tweets. When too many try for the 50K slots to perfectly interleave, stuff happens and TPS drops. Let’s see a few as-measured curves.
1
4
yup fair!
FWIW, my guess is that the tooling around Solana is more likely to break at desired load than the blockchain itself -- RPC servers, explorer, exchange nodes, etc.
(Not sure about that! But those have often been the weakest links in the past.)
1
1
But really -- your core objection is basically the right objection which is "ok sure but in practice will things actually hold up?"
And the only real way to test it is to build it -- so let's build it and see what happens under load!
1
4
(under the realistic load of lots of different people sending messages at random but correlated times that is)
Scaling writes vs scaling reads are two different problems. write optimizing the chain, means indexing is expensive. The cost of indexing 70m non overlapping entries > cost of writing them. Which is fine imho, as long as everyone has access.
6