PHASE 12 Scaling · ~3 hours

Layer 2 & Scaling

Ethereum L1 does ~15 tx/s. That's the global throughput. L2s are how we get to Visa-scale without abandoning L1 security.

Goal — know the difference between optimistic and zk rollups, sidechains, validiums, and deploy your existing contract to an L2.

1. The scaling problem in one sentence

Every full node must re-execute every tx. To scale, you either (a) move execution off L1 and post only results ("rollups"), or (b) run a separate chain ("sidechain") and hope it stays honest.

2. The scaling landscape

Type	Who verifies?	Security	Examples
Optimistic rollup	L1 accepts unless fraud proof within 7d	~L1 (assumes 1 honest watcher)	Optimism, Arbitrum, Base
zk rollup	L1 verifies a succinct proof per batch	L1 (cryptographic)	zkSync, Starknet, Polygon zkEVM
Validium	Like zk but data off-chain	Depends on DA committee	Immutable X
Sidechain	Own validator set	Own security budget	Polygon PoS, BNB Chain
State channel	Peer-to-peer, settle later	L1 for disputes	Lightning (BTC), (rare on ETH)

Analogy — L1 is mainnet Git. L2 is a git-svn clone that does fast local work, then batches "pushes" back. Optimistic = "assume my push is good unless someone cries fraud in a week". zk = "here's a mathematical proof the push is correct".

3. How a rollup works (big picture)

Users send txs to L2 sequencer (fast, cheap) │ ▼ Sequencer batches ~1000 txs, compresses, posts to L1 as calldata/blobs │ ▼ Optimistic: 7-day challenge window to submit fraud proof zk: Prover generates SNARK/STARK, verifier contract checks it on L1 │ ▼ L1 state root updates → withdrawals finalize

The security model is: "L1 is your settlement and data availability layer; the L2 is just an execution shortcut."

4. Costs & UX

Post-EIP-4844 (blobs): L2 tx costs dropped 10–100×. Typical tx on Base/Optimism: $0.01–$0.10.
Withdrawals: optimistic = 7 days (or bridge with liquidity for instant UX). zk = minutes-hours depending on proof cadence.
Same EVM, same Solidity, same wallet, different chainId. Porting is often a config change.

5. Deploying to Base / Optimism / Polygon

// hardhat.config.js
networks: {
  base:    { url: "https://mainnet.base.org", accounts: [PK] },
  optimism:{ url: "https://mainnet.optimism.io", accounts: [PK] },
  polygon: { url: "https://polygon-rpc.com", accounts: [PK] },
}

npx hardhat run scripts/deploy.js --network base

That's it. Verify on the L2's block explorer (Basescan, Arbiscan, etc.) with the same plugin.

6. Nuances that bite in production

Pre-compiles differ slightly — some L2s lack blockhash history or have subtly different block.timestamp.
Gas pricing model differs — Arbitrum has an L1 data fee component; your estimates need to include it.
Sequencer is centralized today — downtime = no tx inclusion. Decentralized sequencers are coming (2025–26).
Bridging — users bridge ETH → L2 ETH. Use canonical bridges for safety; 3rd-party bridges (LayerZero, Wormhole) for speed. Biggest DeFi hacks in history have been bridges.
Chain-agnostic code — use block.chainid not hard-coded IDs. Keep deployments scriptable.

7. Data availability (DA)

"Where is the full tx data stored?" On L1 calldata (rollup) = max security. On a DA committee (validium) = cheaper, weaker. Modular chains (Celestia, EigenDA) provide cheap DA for L2s — this is 2025's hot infra category.

8. zk proofs, intuition-only (Phase 13 goes deeper)

A SNARK lets a prover convince a verifier that "I executed these 1,000 txs correctly starting from state S_0 and ending at state S_1" with a ~200-byte proof verifiable in milliseconds. The L1 verifier contract is just some elliptic-curve pairings.

9. Bridging pattern you'll actually build

// L1 deposit
Bridge.depositETH{value: 1 ether}()  → emits Deposit(user, 1 ether)
// L2 relayer / bridge watcher mints 1 L2-ETH to user on L2
// L2 withdraw
Bridge.withdrawETH(1 ether) on L2     → state posts to L1
// After finality window, user calls Bridge.finalize(user, 1 ether) on L1

Golden rule — never write your own bridge. Use the canonical one or a battle-tested protocol. Bridges are the #1 attack target ($2B+ lost historically).

10. Project

Deliverable — deploy your Phase 6 ERC-20 to Base Sepolia (testnet). Use the canonical bridge UI to move test ETH from Sepolia L1 to Base. Interact from your Phase 7 React app on both chains and handle wallet_switchEthereumChain.

Quiz

Q. On an optimistic rollup, you withdraw funds. Why does the canonical bridge take 7 days?

Validators are slow
L1 needs 7 days to finalize
The challenge period: anyone can submit a fraud proof during that window if the withdrawal references invalid state
Gas fees need to be low

The whole optimistic-security model rests on the challenge window. Shortcut it via a liquidity bridge (3rd party lends you L1 ETH, charges ~0.1%, waits 7d for reimbursement).