A quick recap of the short and medium term of Ethereum scaling.
TLDR:
1. Ultra-high scaling with sharding + rollups will be possible *in phase 1*
2. Sharding is NOT "cancelled"
3. Get on a rollup asap; you get 100x scaling even without eth2
TLDR:
1. Ultra-high scaling with sharding + rollups will be possible *in phase 1*
2. Sharding is NOT "cancelled"
3. Get on a rollup asap; you get 100x scaling even without eth2
The original ETH2 roadmap was created with 3 phases:
0. PoS (this is the one that's coming very soon)
1. Sharding of data, but not of computation (that is, the sharded chain will *include* ~2 MB/sec of data, but it will just be dumb data blobs, not txs)
2. Sharded tx processing
0. PoS (this is the one that's coming very soon)
1. Sharding of data, but not of computation (that is, the sharded chain will *include* ~2 MB/sec of data, but it will just be dumb data blobs, not txs)
2. Sharded tx processing
Currently, we have ~15-45 TPS. Rollups offer a ~100x increase in throughput. Sharding offers a ~64x increase. These two stack multiplicatively; rollups *on top of* sharding offer a ~6400x (!!) increase in throughput.
But the roadmap has an interesting unintended artefact: sharded applications by themselves need phase 2. But sharded rollups only need phase 1, because rollups use the chain only for data, not for computation. So we will have all the tools we need for 6400x throughput quite soon!
So it's not "rollups instead of sharding", it's "rollups on top of sharding". That said, rollups are already here or coming soon even before sharding, and rollups without sharding still offer that 100x increase in throughput. So get on a rollup today!
PS. @lightclients has a compelling take that rollups and sharding-of-tx-execution are actually *very similar* mechanisms. But this gets into more complicated theory and is further away from concrete consequences to applications...
lightclient.io/blog/eth2-is-a…
lightclient.io/blog/eth2-is-a…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
