Content
@
0 reply
0 recast
0 reaction
Marius van der Wijden
@vdwijden
I recently wrote down some ideas on how to built a high throughput L2. Some of the design decisions that I would take (and that I suspect other rollups take as well) and why they can't be done on mainnet Ethereum. Check it out! https://mariusvanderwijden.github.io/blog/2024/07/12/L2s/
2 replies
6 recasts
29 reactions
Sergey Potekhin
@fastfourier.eth
Great article! > In my opinion, this way of parallelizing computation can only help the average case Do you think it's possible to effectively stimulate to use "sharded" architecture for dapps on such chains?
1 reply
0 recast
0 reaction
Marius van der Wijden
@vdwijden
I think so, by making parallelizable transactions cheaper, you can encourage developers to think about sharded designs in order to stay competitive with others. Once the first dapp uses a sharded design, others will follow to provide the cheapest service for their users (similar to the gas golfing we're seeing on mainnet)
1 reply
0 recast
0 reaction
Sergey Potekhin
@fastfourier.eth
Agree. Based on my experience, I believe it's not just about making the 'right' design cheaper to use but also about making the 'wrong' design significantly more costly. Hard to express in precise numbers, but intuitively, a 'wrong' design causes more harm to a sharded architecture than a 'right' design supports it. In other words, while 100 dApps with a sharded design might keep the system running smoothly, a single large ERC20 USDT contract easily impacts the performance of the entire network
0 reply
0 recast
1 reaction
