3 ms·

> So they are breaking down the blockchain into small Validator groups. You have to know which Cluster you want to contact. That's... a compromise. It's not eve

> So they are breaking down the blockchain into small Validator groups. You have to know which Cluster you want to contact. That's... a compromise. It's not even comparable with ETH or the internet where you have this massive pool of validators that all participate over the same network. No, the thing you quoted just says that devnet validators ignore transactions intended for mainnet and testnet. > Ok, so, if I have a cluster where I know all the Validators, I can collude with them to just ignore any new Validators from coming in? Yes, but "a cluster" is "the whole blockchain." This is analogous to saying that you know every Bitcoin miner or every Ethereum node. > If I'm the leader, can I censor a transaction coming from a Validator? Or perhaps I see a transaction that will move the market, can I sneak in a transaction in my own benefit before broadcasting that one to my validators? If you are a leader, you can censor transactions, and there will be another leader 2 seconds later. This is similar to the situation in literally every blockchain. In Ethereum today, miners are inserting transactions to extract value from traders. You can read about this here: http://mev.wiki/ http://mev.wiki/

a1371 5y ago
Good comment on testnet. From what I understand, I don't think that's all the plan for the these clusters. Otherwise, why do they specify this? > When two clusters share a common genesis block, they attempt to converge. More importantly, the cluster is only 150 validators: > A Solana cluster is capable of subsecond confirmation for up to 150 nodes with plans to scale up to hundreds of thousands of nodes. This is not comparable to Mempool where it is in your best interest to broadcast a transaction across the network. Yes, everyone sees the mempool but it doesn't allow you to manipulate transaction order as long as the original sender has paid a high enough fee.
- anonymoushn 5y ago
  > > When two clusters share a common genesis block, they attempt to converge. This is a situation analogous to conflicting chains during a bitcoin block reorg, made more complex by the fact that validators have opinions about which other validators they are in a cluster with. > More importantly, the cluster is only 150 validators: > > A Solana cluster is capable of subsecond confirmation for up to 150 nodes with plans to scale up to hundreds of thousands of nodes. It's like 700 currently. Larger validator sets lead to logarithmically slower confirmations. At the time that doc was written, one point on that curve was (150 validators, 1 second) or so.

a1371 5y ago

Good comment on testnet. From what I understand, I don't think that's all the plan for the these clusters. Otherwise, why do they specify this? > When two clusters share a common genesis block, they attempt to converge. More importantly, the cluster is only 150 validators: > A Solana cluster is capable of subsecond confirmation for up to 150 nodes with plans to scale up to hundreds of thousands of nodes. This is not comparable to Mempool where it is in your best interest to broadcast a transaction across the network. Yes, everyone sees the mempool but it doesn't allow you to manipulate transaction order as long as the original sender has paid a high enough fee.

anonymoushn 5y ago
> > When two clusters share a common genesis block, they attempt to converge. This is a situation analogous to conflicting chains during a bitcoin block reorg, made more complex by the fact that validators have opinions about which other validators they are in a cluster with. > More importantly, the cluster is only 150 validators: > > A Solana cluster is capable of subsecond confirmation for up to 150 nodes with plans to scale up to hundreds of thousands of nodes. It's like 700 currently. Larger validator sets lead to logarithmically slower confirmations. At the time that doc was written, one point on that curve was (150 validators, 1 second) or so.