> Where are you getting ideas like this? How fast do you think internet connections are?
As fast as the packet source wants? The only way to defend against a slowloris DoS is by accounting for it at the application layer, which is both unusual and difficult - as DoS attacks are usually handled at the transport layer. In this case that could mean applying a deadline for the block announcement in its entirety - which means mandating a minimum connection speed, and classifying every connection that dips below that as an attack... so you better have a good SLA with your upstream network. Well, that is no problem for centralized operations... Guess what happens to cryptocurrencies that adopt massive 128MB+ blocks in order to increase their transaction throughput - they become incredibly fragile as the slightest amount of packet loss starts waves of peer bans.
> Who are these miners that are struggling to send out 900KB ?
Anyone who wants to handicap a competing pool by forcing them to either wait for the complete false block, or waste time hashing on a fake block header.
When are you going to admit that you haven't thought this through?
As fast as the packet source wants? The only way to defend against a slowloris DoS is by accounting for it at the application layer, which is both unusual and difficult - as DoS attacks are usually handled at the transport layer. In this case that could mean applying a deadline for the block announcement in its entirety - which means mandating a minimum connection speed, and classifying every connection that dips below that as an attack... so you better have a good SLA with your upstream network. Well, that is no problem for centralized operations... Guess what happens to cryptocurrencies that adopt massive 128MB+ blocks in order to increase their transaction throughput - they become incredibly fragile as the slightest amount of packet loss starts waves of peer bans.
> Who are these miners that are struggling to send out 900KB ?
Anyone who wants to handicap a competing pool by forcing them to either wait for the complete false block, or waste time hashing on a fake block header.
When are you going to admit that you haven't thought this through?