Layer 2 Scaling

Module 1 of Advanced Protocols

The Scaling Problem

Ethereum processes ~15 transactions per second. Visa processes ~65,000. This limitation comes from blockchain's core trade-off: every node must verify every transaction.

The Trilemma

         Decentralization
              ╱╲
             ╱  ╲
            ╱    ╲
           ╱  ??  ╲
          ╱________╲
    Security ────── Scalability

Traditional blockchains sacrifice scalability. Layer 2s try to have all three.

How L2s Work

Core Insight: Execute transactions off-chain, inherit L1 security.

Layer 2 (Fast, Cheap)          Layer 1 (Secure, Slow)
┌──────────────────┐           ┌──────────────────┐
│  Tx1, Tx2, Tx3   │           │                  │
│  Tx4, Tx5, Tx6   │──Batch───▶│  Proof/Data      │
│  Tx7, Tx8, Tx9   │           │  Posted          │
└──────────────────┘           └──────────────────┘
   1000s of TPS                   15 TPS
   $0.01/tx                       $10/tx

L2s batch many transactions and post a summary to L1. Security comes from the ability to verify this summary on L1.

Rollup Types

Optimistic Rollups

Philosophy: Assume transactions are valid, challenge if fraud detected.

Transaction Flow:
1. Batch posted to L1
2. 7-day challenge period begins
3. Anyone can submit fraud proof if invalid
4. No challenge? Batch finalized

┌─────────────────────────────────────────┐
│  Batch Posted  │  Challenge Window  │ Final
└────────────────┴─────────────────────┘
     Day 0              Days 1-7        Day 7+

Pros:

• EVM compatible (easy to deploy existing code)
• Lower computation overhead
• Mature ecosystem (Arbitrum, Optimism, Base)

Cons:

• 7-day withdrawal delay
• Requires active fraud watchers
• Data still on-chain

ZK Rollups

Philosophy: Prove validity mathematically, no trust required.

Transaction Flow:
1. Execute transactions off-chain
2. Generate cryptographic proof
3. Post proof + data to L1
4. L1 verifies proof → immediately valid

┌─────────────────────────────────────────┐
│  Execute  │  Prove  │  Verify  │ Final
└───────────┴─────────┴──────────┘
   Fast        Slow      Instant    Instant

Pros:

• Instant finality (no challenge period)
• Stronger security guarantees
• More data compression possible

Cons:

• Expensive proof generation
• EVM compatibility harder (zkEVMs)
• Less mature

Major L2 Comparison

Values approximate and fluctuate

Data Availability

Rollups must make transaction data available so anyone can reconstruct state and verify/challenge.

Options

1. On-Chain (Rollup)

• All data posted to L1
• Most expensive but most secure
• Used by: Arbitrum, Optimism, zkSync

2. Off-Chain (Validium)

• Data stored elsewhere (committee, DA layer)
• Cheaper but additional trust assumptions
• Used by: Immutable X, some Starknet modes

3. Hybrid (Volition)

• Users choose per-transaction
• Security vs cost trade-off per use case

EIP-4844 (Proto-Danksharding)

Introduced "blob" transactions:

• Temporary data storage (~18 days)
• 10-100x cheaper than calldata
• Rollups post data in blobs
• Reduced L2 fees significantly

Bridging to L2

Deposit (L1 → L2)

1. Lock funds on L1 bridge contract
2. L2 sees deposit transaction
3. Mint equivalent tokens on L2
4. User can transact on L2

Time: ~10 minutes (depends on L2)

Withdrawal (L2 → L1)

Optimistic Rollups:

1. Initiate withdrawal on L2
2. Wait 7 days challenge period
3. Prove withdrawal on L1
4. Claim funds

Time: 7+ days (can use fast bridges)

ZK Rollups:

1. Initiate withdrawal on L2
2. Wait for next proof submission
3. Claim funds on L1

Time: Hours (depends on proof frequency)

Fast Bridges

Services that front liquidity to avoid withdrawal delays:

• Hop Protocol
• Across
• Stargate

User pays small fee, receives funds quickly.

zkEVM Types

Not all ZK rollups are created equal:

Trade-off: More compatibility = slower proofs

• Polygon zkEVM: Type 2 (high compatibility)
• zkSync Era: Type 4 (custom compiler)
• Starknet: Type 4 (Cairo language)

L3s and Appchains

Layer 3s

L2s on top of L2s:

L1 (Ethereum)
    └── L2 (Arbitrum)
            └── L3 (Your App Chain)

Use cases:

• Application-specific chains
• Custom execution environments
• Further cost reduction

Appchains

Dedicated chains for single applications:

• Custom parameters
• Dedicated blockspace
• Can use any DA layer

Examples:

• dYdX (moved from L2 to Cosmos appchain)
• Immutable X (gaming-focused)

Sequencers and Decentralization

Current State

Most L2s have centralized sequencers:

• Single entity orders transactions
• Can censor transactions
• Can extract MEV
• Single point of failure

Decentralization Roadmap

Current           Near-term          Long-term
Centralized  →   Permissioned   →   Fully
Sequencer        Sequencer Set      Decentralized

Shared Sequencing

Multiple L2s share decentralized sequencer:

• Atomic cross-L2 transactions
• Reduced fragmentation
• Network effects

Security Considerations

Optimistic Rollup Risks

• Sequencer down: L1 escape hatch exists
• No challengers: Need at least one honest watcher
• Long delays: 7 days for withdrawals

ZK Rollup Risks

• Prover centralization: Most have centralized provers
• Upgrade risk: Many have upgradeable contracts
• New cryptography: Less battle-tested

What to Check

Before using an L2:

1. Is there an L1 escape hatch?
2. Who can upgrade contracts?
3. How decentralized is sequencing?
4. What's the challenge/proof mechanism?

The L2 Landscape

                    L2 Ecosystem
    ┌─────────────────────────────────────────┐
    │         Optimistic Rollups              │
    │  ┌──────────┐ ┌──────────┐ ┌────────┐   │
    │  │ Arbitrum │ │ Optimism │ │  Base  │   │
    │  └──────────┘ └──────────┘ └────────┘   │
    ├─────────────────────────────────────────┤
    │            ZK Rollups                   │
    │  ┌──────────┐ ┌──────────┐ ┌────────┐   │
    │  │  zkSync  │ │ Starknet │ │  Scroll│   │
    │  └──────────┘ └──────────┘ └────────┘   │
    └─────────────────────────────────────────┘
                        │
                   All settle to
                        │
                        ▼
                ┌───────────────┐
                │   Ethereum    │
                │   Layer 1     │
                └───────────────┘

Key Takeaways

1. L2s inherit L1 security while providing cheap, fast transactions
2. Optimistic rollups are mature and EVM-compatible but have withdrawal delays
3. ZK rollups offer faster finality but are more complex
4. Data availability is the key trade-off (cost vs security)
5. Sequencer decentralization is still in progress
6. EIP-4844 dramatically reduced L2 costs
7. The future is multi-L2 with shared sequencing and interoperability