1) Why rollups?

• Cost: Bundle many transactions into one L1 transaction → users pay a fraction of L1 gas.
• Throughput: Parallel execution + compressed posting to L1 → higher TPS.
• Security: Dispute/validity proofs on L1 inherit Ethereum’s security assumptions.

2) How rollups work (high level)

1. Users send txs to an L2 node/sequencer.
2. Sequencer orders and executes txs off-L1, producing a new L2 state root.
3. Commitment (batch) and data are posted to L1 for availability.
4. Proof (fraud or validity) lets L1 verify correctness or allow challenges.

3) Optimistic rollups & fraud proofs

Optimistic rollups (e.g., Arbitrum, Optimism, Base) assume batches are valid when posted to L1.
A challenge window (typically minutes to days) lets anyone submit a fraud proof
showing incorrect execution. If a batch is proven invalid, it’s reverted/slashed.

• Pros: Simple prover side; mature EVM compatibility; good dev tooling.
• Cons: Withdrawals to L1 are delayed until the challenge window expires.
• Security: Requires at least one honest challenger with data access.

4) ZK rollups & validity proofs

ZK rollups (e.g., zkSync, Starknet, Scroll, Polygon zkEVM) generate a cryptographic validity proof
that attests a batch was executed correctly. L1 verifies the proof in milliseconds no challenge window.

• Pros: Fast finality & withdrawals; censorship-resistant proving assumptions.
• Cons: Provers are complex/expensive; some L2s are still converging on full EVM equivalence.
• Security: As long as proofs verify and data is available, state is final on acceptance.

5) Data Availability (DA): L1, Blobs, Alt-DA

To reconstruct and challenge L2 state, transaction data must be accessible (available) somewhere durable.

• Post to L1 calldata: Max security, higher cost.
• Blobspace (EIP-4844): Cheaper ephemeral blobs on Ethereum for rollup data → big fee reductions.
• Alt-DA layers: Specialized networks (e.g., Celestia, EigenDA) reduce costs but add new trust/assumptions.

6) Bridging & withdrawal times

• Optimistic → L1: Wait the challenge window (hours–days) for native bridge withdrawals.
• ZK → L1: Withdrawals can finalize within minutes once proofs are verified.
• Third-party fast bridges: Provide liquidity to skip waits; introduce bridge smart-contract risk.

7) Risks & decentralization roadmaps

• Sequencer centralization: Many rollups run a single sequencer; decentralization/MEV roadmaps are in progress.
• Admin keys: Upgrades/pauses controlled by multisigs at early stages; check timelocks & governance.
• Prover/Verifier bugs: Cryptography is hard, audits and battle testing matter.
• DA assumptions: Alt-DA requires trust in that network’s availability/finality.

8) Picking an L2 for your app

Further resources

• Ethereum.org  Rollups overview
• OP Stack (Optimism), Arbitrum Nitro  architecture docs
• zkSync, Starknet, Scroll, Polygon zkEVM  prover/zkEVM docs
• EIP-4844 (Proto-Danksharding)  blobspace for cheaper L2 data
• Celestia / EigenDA  alternative data availability networks
• Want to go deeper into dev? Try Cyfrin Updraft for end-to-end smart contract courses.