Blog

← Writing

Verkle Trees

May 14, 2021
Coming soon...Heres a preview

State in Ethereum is not a flat parcel of land - its uneven, rocky, and brittle.
Merkle trees are cool, but when you think of it

  1. All major retail / normie crypto activity takes place over metamask and now things like Rainbow.me / Coinbase Wallet (#9 on app store!)
  2. They typically use infura / alchemy ~ infrastructure provider.

State availability is a new interesting problem I've started keeping track of.

The first step? Need Look into Verkle Trees

The gist:
Verkle trees seem to be a bandwidth-efficient alternative to merkle trees. This is cool, and might reduce the overhead for p2p gossip over ethereum (I think RLPx has a message size limit of ~16 Mb, and most ethereum related p2p messages are around 10Mb)
Theres some talk about a 'Portal Network' within ETH R&D and its really cool, probably because it partially relates to devp2p, something which I've been interested in for a bit.

Edit: The portal network now has a nice readme here.

Status is building an implementation called Fluffy, I found a typescript implementation (seems to be WIP) here. I'm closely watching trin, which is written in rust.

Verkle trees reduce the bandwidth required to move data around. Do the math. Can there exist a decentralized infura alongside a live, running version of eth? Theres p2p already, right?

Could it be possible that in the future, 100k-200k low-resource clients (e.g: 128 Mb memory) are quietly helping core eth p2p layer?

It is possible the people will share state out of pure altruism, one of the core pillars in the ETH community.
This changes the game - UX, onboarding, surface area for new apps..

Will discuss more after finals.

Subscribe to the newsletter

Get updates about new posts, new projects, or other meaningful updates to this site delivered to your inbox. Alternatively, you can follow me on Twitter.