A quick and messy intro to a huge and equally messy topic

Layer 2

Sharding

Heterogeneous Sharding

A few of buzzwords I won't be able to explain adequately:

- ZKSnarks (friday talk soon™)

- Lightning Networks

- Cross-chain transactions

- Verifiably Random Functions (VRFs)

Layer 2 Protocols

Optimistic Rollups

ZK Rollups

• rollup data is assumed valid (hence the optimism)
   
• Not much off-chain computation needed
   
• A challenge period is needed, during which Fraud Proofs can be submitted
   
• Fully EVM-compatible (i.e.: can run Smart Contracts and Geth nodes)

• All data commited to the main chain is cryptographically validated
   
• ZK proofs are expensive to compute
   
• No challenge period, because math is magical
   
• Smart contract support is trickier

Optimism, Arbitrum

zkSync, Loopring, ...

PS: What is Polygon?

a completely independent and separate blockchain with validators, and a cron job that stores the network state hash on the Ethereum blockchain for shits and giggles.

- Random hacker news comment

Sharding

(Homogeneous, the ETH 2.0 kind)

Single chain (no sharding)

10 shards

• 100 validators, all working on the same state
   
• a 51% attack can occur when 51 validators are ccorrupt, and working together against the other 49

• 100 validators, 10 per shard
   
• cross-shard communication is needed, so each shard needs to trust the others
   
• A single shard can be compromised with 6 corrupt nodes. That then bleeds to the whole system (VFRs can help here)

App-specific

(Heterogeneous Sharding)

Homogeneous

Heterogeneous

• Every shard is equal
  
   
• Cross-shard communication is needed at random. You don't choose which shard you live in
   
• No need to elect shards

• Every shard is unique
  
   
• Cross-shard communication is more focused and predictable (akin to a 3rd party API call)
   
• Each shard is its own product. Voting or auctions are needed to allocate each slot