Types of Layer 2 in Blockchain: Rollups, Plasma & More

The main types of Layer 2 are Rollups (Optimistic and ZK), State Channels, Validium, and Plasma. Each processes transactions off-chain and settles results back to a base blockchain like Ethereum, but they differ in how they prove validity and where they store data.

Those differences determine how fast your funds finalize, how much you pay per transaction, and how much trust you place in the system.

What Is Layer 2 in Blockchain?

Most base-layer blockchains have a throughput ceiling which creates congestion and high fees during peak demand. Layer 2 networks address this by taking most of the computational work off-chain, then anchoring their results back to the base chain's security.

Bitcoin's Lightning Network, Ethereum's Arbitrum and zkSync, and many others all follow this same principle on different underlying chains. 

The key distinction between L2s and independent blockchains is where security comes from. A true L2 relies on the base layer to finalize and protect its state.

What Are the Main Types of Layer 2?

Author's Note: What makes L2 architecture genuinely interesting is that each design is essentially a different answer to the same question: "How much do you trust the parties involved?" Optimistic Rollups assume honesty and punish fraud. ZK Rollups prove correctness mathematically. State Channels require participants to already know each other. Validium trusts a data committee. Plasma trusted child chain operators. The spectrum runs from cryptographic certainty to social trust. Where a design sits on that spectrum determines almost everything about its tradeoffs.

Rollups

Rollups are the dominant Layer 2 design today. They bundle (or "roll up") hundreds to thousands of off-chain transactions into a single batch, then post compressed transaction data and a proof of validity to the Ethereum mainnet.

Unlike earlier L2 designs, rollups publish their transaction data on-chain, which means anyone can reconstruct the rollup's state from Ethereum alone, without trusting any external party.

Rollups come in two primary variants, distinguished by how they prove that batches are valid:

Optimistic Rollups

Optimistic Rollups assume all submitted transactions are valid by default, hence the name. Rather than generating a cryptographic proof for every batch, they post transaction data to Ethereum and wait.

Anyone who spots an invalid transaction during the challenge window (typically 7 days) can submit a fraud proof to contest it. If no challenge is raised, the batch is finalized.

How it works in practice:

1. Transactions are executed off-chain by a sequencer.
2. The batch (with full transaction data) is posted to Ethereum.
3. A 7-day challenge period begins. Watchers re-execute the batch locally and compare results.
4. If a fraud proof is submitted and verified, the invalid batch is rejected and the sequencer is penalized.
5. If no challenge occurs, the state is finalized on L1.

Key tradeoff: The 7-day withdrawal window is a direct consequence of this model. To move assets back to the Ethereum mainnet, users must wait out the full challenge period to ensure no pending fraud proof can invalidate their withdrawal.

Real-world examples:

• Arbitrum One: Largest L2 by TVL, using a multi-round interactive fraud-proof system. As of late 2025, Arbitrum held approximately $16–19B in total value locked.
• Base: Built on the OP Stack by Coinbase, surpassing $10B TVL in 2025.
• OP Mainnet (Optimism): The original Optimistic Rollup, foundation of the Superchain ecosystem.

ZK Rollups

Rather than assuming validity and allowing fraud to be disputed later, ZK Rollups prove the validity of every batch upfront using a cryptographic validity proof – specifically, a zk-SNARK or zk-STARK.

How it works:

1. Transactions are executed off-chain.
2. A cryptographic validity proof is generated that mathematically certifies every state transition in the batch was executed correctly.
3. The proof (plus compressed transaction data) is posted to Ethereum.
4. Ethereum's smart contract verifies the proof. This happens near-instantly.
5. Once verified, the batch is finalized. No waiting period required.

The tradeoff: Generating ZK proofs is computationally intensive. Historically, this made ZK Rollups more expensive to run and harder to make EVM-compatible — but both challenges have been substantially addressed. As of 2025–2026, all major ZK Rollups offer some level of EVM compatibility.

Real-world examples:

• zkSync Era: General-purpose ZK Rollup with full EVM compatibility, launched March 2023.
• Starknet: Uses STARK proofs (no trusted setup requirement), with its own Cairo smart contract language.
• Scroll and Polygon zkEVM: Type-2 zkEVMs, prioritizing near-perfect Ethereum equivalence.

State Channels

State Channels are the oldest form of Layer 2 design. They allow two or more participants to transact freely off-chain through a private "channel," only touching the blockchain twice: once to open the channel (locking funds into a smart contract) and once to close it (settling the final balance).

How it works:

1. Participants deposit funds into a shared smart contract on Ethereum, opening the channel.
2. They exchange signed off-chain messages representing updated balances – as many times as needed, at zero cost.
3. When done, either party broadcasts the final signed state to Ethereum. The contract distributes funds accordingly.
4. If one party tries to submit an outdated state to cheat, the other can submit a more recent signed state to override it during a dispute window.

What they're good at: State Channels excel at scenarios involving many rapid transactions between known parties – micropayments, gaming turns, or any high-frequency back-and-forth where paying gas per transaction is impractical.

Limitations: Channels only work between participants who have opened a channel together. They cannot support complex smart contracts or interactions with arbitrary third parties, which limits their applicability for general DeFi use.

Real-world examples:

• Lightning Network (Bitcoin layer 2): The most widely deployed state channel network, enabling fast, low-cost Bitcoin payments.
• Raiden Network (Ethereum): The Ethereum equivalent of Lightning.

Validium

Validium is closely related to ZK Rollups. It uses the same cryptographic validity proofs to confirm that every transaction batch is correct. The key difference is where transaction data is stored.

In a ZK Rollup, full transaction data is posted to Ethereum L1 (ensuring anyone can reconstruct the state independently). In a Validium, transaction data is kept off-chain, managed by a trusted Data Availability Committee (DAC) – a group of designated entities that store the data and attest to its availability.

This off-chain data strategy eliminates L1 data publishing costs, enabling throughputs in the range of 9,000–10,000+ TPS and sub-cent transaction fees.

The cost: If the DAC goes offline or withholds data, users cannot prove their asset ownership to exit. They would need to rely on the committee to cooperate.

Real-world examples:

• StarkEx (StarkWare): Powers Immutable X (NFT gaming, gas-free minting) and previously dYdX v3.
• Immutable X: Processes NFT transactions at scale with no gas fees for users.

Plasma (Legacy Layer 2 Design)

Plasma was introduced in 2017 by Vitalik Buterin and Joseph Poon as one of the first serious Layer 2 scaling proposals for Ethereum. It works by creating child chains – smaller blockchains that:

• Run alongside Ethereum
• Process their own transactions
• Periodically commit state roots back to the mainchain using fraud proofs

The concept was architecturally elegant: child chains handle volume, Ethereum handles security. In practice, Plasma ran into a fundamental challenge known as the data availability problem.

If the operator of a child chain withholds transaction data, honest users have no way to construct fraud proofs to challenge invalid exits. The primary safety mechanism depends on having access to chain data that users might not reliably be able to obtain.

Historical examples:

• OMG Network (formerly OmiseGO): One of the earliest Plasma deployments.
• Polygon PoS: Originally incorporated Plasma-style security mechanisms for its bridge, though it operates as a sidechain, not a pure L2. For a detailed breakdown of that distinction, see Sidechain vs Layer 2: What's the Difference?

Current status: Plasma is largely considered a legacy design. It clarified the data availability problem and influenced the architecture of modern rollups, but it is no longer the basis for new L2 development.

Why Do Different Layer 2 Types Exist?

The core tension is between data availability and cost:

• Posting full transaction data to Ethereum (as rollups do) is the most secure approach. Anyone can verify the chain state independently. But it costs more, because Ethereum blockspace is finite.
• Keeping data off-chain (as Validium does) is cheaper and faster, but introduces trust assumptions about whoever holds the data.
• Using fraud proofs (Optimistic Rollups) delays finality but avoids expensive proof generation. Using validity proofs (ZK Rollups) gives instant cryptographic certainty but requires significant computation.

This is why the ecosystem supports multiple L2 types. Different applications have genuinely different requirements. A decentralized exchange handling $10M settlements has different security needs than a blockchain game processing millions of small in-game transactions.

As ethereum.org puts it: "Different solutions can exist and work in harmony, allowing for an exponential effect on future transaction speed and throughput."

Layer 2 Types Compared: Quick Reference Table

No two L2 types make the same tradeoffs. The table below maps each design across the four dimensions that matter most: how it proves validity, whether data lives on L1, how fast it finalizes, and how compatible it is with Ethereum's tooling.

The two columns that reveal the most are Data on L1 and Finality. Together, they show exactly what each type trades away to get its performance gains.

>> Read more: Layer 1 vs Layer 2: The Evolution of Blockchain Scaling

Which Type of Layer 2 Should You Choose?

Best for general Ethereum scaling

→ Optimistic Rollup (Arbitrum, Base, OP Mainnet)

If you're building or using a DeFi protocol, deploying existing Solidity contracts, or want access to deep liquidity and a mature ecosystem, Optimistic Rollups are the default choice. They offer near-perfect EVM equivalence, meaning most Ethereum code runs without modification.

By late 2025, Arbitrum and Base collectively controlled roughly 80% of all L2 TVL – a signal of where developers and liquidity have concentrated.

The 7-day withdrawal window is a real constraint but rarely a practical problem for most users, since fast bridges can provide instant exits for a small fee.

Best for the lowest transaction costs

→ Validium (Immutable X, StarkEx-based platforms)

For applications that need to process large volumes of low-value transactions, including NFT minting, in-game item transfers, and micro-payments, Validium offers the lowest cost per transaction.

By keeping data off-chain, Validium platforms can reach 9,000–10,000+ TPS at sub-cent fees, compared to the higher (though still inexpensive) costs of ZK Rollups.

The tradeoff is accepting the DAC trust model.

• For gaming NFTs or casual trading, Validium is generally reasonable.
• For high-value asset storage, a full ZK Rollup is a safer choice.

Best for high-throughput applications

→ ZK Rollup or Validium, depending on data availability requirements

ZK Rollups like Starknet and zkSync Era offer high throughput with full data availability on Ethereum, meaning no trust in a committee, just math. For applications needing both scale and the highest security guarantees (institutional settlement, payment systems), ZK Rollups are the correct tradeoff.

For even higher raw throughput, where some trust in a data committee is acceptable, Validium extends the ZK Rollup architecture with off-chain data for extreme scale.

Best for security-critical use cases

→ ZK Rollup

When transaction finality and cryptographic certainty matter most, like bridges moving large assets, protocols handling institutional capital, or any system where the cost of a single error is catastrophic, ZK Rollups are the correct choice.

The validity proof model guarantees that only valid state transitions can be posted to Ethereum. There is no optimistic window during which an invalid batch might slip through. Withdrawals finalize in hours, not days, and the security model relies entirely on mathematics rather than economic incentives or social assumptions.

Sources and Further Reading

• Ethereum L2 Scaling Documentation – ethereum.org
• Optimistic Rollups – ethereum.org
• Data Availability in Layer 2 Rollups: A Survey (MDPI, 2024)
• ZKsync Finality Documentation – docs.zksync.io
• Plasma Whitepaper – Buterin & Poon (2017)
• State Channels – ethereum.org