Layer 2 networks compared
Ethereum’s base layer can process roughly 15 transactions per second. That is not enough for widespread use, and the cost of each transaction rises when demand spikes. Layer 2 networks solve this by moving execution off the main chain while inheriting some or all of its security guarantees. But “Layer 2” covers several distinct architectures, each with different tradeoffs in speed, cost, safety, and developer experience. This page maps the entire landscape: what the major designs are, which tools exist today, and - most importantly - which spoke page answers the specific questions you are likely to have.
The two main architectures: optimistic rollups and ZK-rollups
Every L2 network can be classified by how it proves that its off-chain execution was correct. There are two families: optimistic rollups, which assume transactions are valid unless challenged, and ZK-rollups, which generate cryptographic proofs of correctness for every batch.
Optimistic rollups post transaction data to Ethereum and wait through a challenge period - typically seven days - during which anyone can submit a fraud proof to contest an invalid state transition. If no one challenges, the batch is confirmed. The most widely used optimistic rollup chains are Arbitrum One and Optimism OP Mainnet. Both use fraud proofs, but they implement them differently: Arbitrum uses interactive dispute resolution through its Nitro architecture, while Optimism originally used a single-round fault proof system and has since moved to a more modular approach with its OP Stack. The dedicated comparison page Arbitrum vs Optimism Which Should I Deploy On breaks down these differences for anyone choosing a chain to build on.
ZK-rollups take the opposite approach. They compress thousands of transactions into a single batch, generate a validity proof (either a SNARK or a STARK), and post that proof to Ethereum alongside the state commitment. Because the proof is verified on-chain, there is no waiting period - the batch is final as soon as Ethereum confirms it. Major ZK-rollup chains include zkSync Era, Starknet, Polygon zkEVM, Scroll, and Linea. The tradeoff is that generating these proofs is computationally expensive, and not all ZK-rollups are equally compatible with existing Ethereum tooling. The page ZK Rollup vs Optimistic Rollup for Institutional Use examines which architecture suits regulated entities that need predictable finality and provable correctness.
A common belief is that ZK-rollups offer instant withdrawals today. In practice, the withdrawal process still involves posting a proof to L1 and waiting for confirmation, plus the proving system itself may introduce delays of minutes to hours. The page Do ZK Rollups Actually Have Instant Withdrawals Today addresses this misconception directly.
The major L2s: which one fits your use case
Choosing a specific chain matters now more than ever. The ecosystem is not a single L2 but a collection of competing networks, each with different governance, fee models, and risk profiles.
For general-purpose applications, Arbitrum One and OP Mainnet are the incumbents. Arbitrum has a larger total value locked and a more proven track record of uptime. Optimism pioneered the Superchain concept - a network of interoperable chains built on the shared OP Stack. The page Arbitrum vs Optimism Which Should I Deploy On walks through the concrete differences in gas costs, developer tooling, and withdrawal times.
Base, incubated by Coinbase, is the fastest-growing OP Stack chain. It benefits from Coinbase’s user base and distribution but introduces centralization concerns. The page Base vs OP Mainnet Inside the Superchain Ecosystem answers whether deploying on Base versus OP Mainnet changes your security assumptions, and the dedicated reality-check page Is Base Fully Decentralized Now Reality Check explains what centralized control the Base sequencer retains today.
Polygon PoS is frequently called a Layer 2 but is architecturally a sidechain with its own consensus mechanism. It does not post transaction data to Ethereum, meaning it does not inherit Ethereum’s security. The page Polygon PoS vs Arbitrum One What is the Difference makes this distinction explicit and explains why one is a rollup and the other is not.
For ZK-rollup developers, the choice between zkSync Era and Polygon zkEVM often comes down to EVM compatibility. zkSync Era uses a custom virtual machine that is compatible with Solidity but not bytecode-identical to the EVM, while Polygon zkEVM aims for full bytecode equivalence. The page Polygon zkEVM vs zkSync Era for EVM Compatibility lists the concrete differences in deployment friction. Meanwhile, Starknet uses a STARK-based proving system with its own Cairo language, while Scroll uses a SNARK-based prover that targets EVM bytecode equivalence. The page Starknet vs Scroll Validity Proof Type Differences explains how the choice of proving system affects developer onboarding and finality time.
For gas-sensitive applications like gaming, Arbitrum Nova offers a tradeoff: it uses an AnyTrust data committee model that stores transaction data off-chain with a trusted committee rather than posting it all to Ethereum. This lowers costs significantly but introduces a new trust assumption. The page Arbitrum One vs Arbitrum Nova for Gaming and Gas Costs helps developers decide when the cost savings justify the reduced security model.
Immutable X uses StarkEx as a validium for gaming and NFTs, keeping data off-chain entirely. The page on Validium vs Volition tradeoffs, covered in the context of the Immutable X ecosystem, explains this distinction.
How to move assets: native bridges versus third-party protocols
Every L2 has a native bridge that locks tokens on Ethereum and mints equivalents on the L2. These bridges are operated by the same entity that controls the L2’s sequencer or governance. They are generally considered the most secure option for moving assets, but they are slow: optimistic rollup withdrawals require a full seven-day fraud proof window, and even ZK-rollup withdrawals require L1 confirmation of the proof.
The page Native Bridge vs Third Party Bridge for Moving Assets compares this approach to using protocols like Across Protocol, Hop Protocol, and Orbiter Finance. Third-party bridges often front liquidity to give users near-instant finality, but they introduce additional trust in the bridge’s operators and smart contracts. The page explains when the speed advantage outweighs the extra risk.
A common frustration occurs when a bridge transaction shows as “pending” for the full fraud proof window. The page Bridge Stuck Pending Fraud Proof Window 7 Day Delay explains why this happens, whether you can speed it up, and what to do if the bridge UI fails to display progress correctly.
If the L2 sequencer goes offline, most rollups have a forced inclusion mechanism that lets users submit transactions directly to the L1 bridge contract. The page Sequencer Offline Mode Forced L1 Withdrawal Path describes how to use this escape hatch, what it costs, and why it may not work as smoothly as users expect.
The frameworks for launching your own L2
You do not have to choose an existing L2. You can launch your own using an L2 framework. The two major options are OP Stack (from Optimism) and Arbitrum Orbit (from Arbitrum). Both provide modular toolkits for deploying a chain that posts data to Ethereum, uses its own sequencer, and integrates with the parent ecosystem.
The page OP Stack vs Arbitrum Orbit for Launching an Appchain compares the two in terms of customization options, governance requirements, and integration with the Superchain or Arbitrum ecosystem. A third option is zkStack from Matter Labs, which targets ZK-rollup hyperchains. The page on L3 Appchain on Arbitrum Orbit vs OP Stack vs zkStack (covered separately) extends this comparison to layer 3s.
A related design choice for rollup developers is the data availability layer. Traditional rollups post all transaction data to Ethereum as calldata or, since the Dencun upgrade, as blobs. EigenDA and Celestia offer alternative data availability layers that reduce costs further but introduce additional trust assumptions. The page Celestia DA vs EigenDA for Rollup Data Availability Layer explains the tradeoffs in latency, cost, and security for developers considering these options.
The cost of posting data to L1 has been the dominant fee on most rollups. The page Blob Transactions vs Calldata for L2 Data Posting Costs breaks down how EIP-4844 changed the unit economics for rollup operators and users, and why blob fees can still spike.
Common risks and misconceptions
The phrase “L2 inherits Ethereum security” is repeated so often that it has lost its precision. In reality, different L2 designs inherit different amounts of Ethereum security. An optimistic rollup that posts all transaction data to Ethereum and allows anyone to challenge state transitions is the closest to full inheritance. A validium that stores data off-chain does not inherit Ethereum’s data availability guarantees, meaning the operator could freeze funds permanently by withholding data. The page Do All L2s Actually Inherit Ethereum Full Security lists each design’s additional trust assumptions in plain terms.
Another widespread belief is that seeing a transaction confirmed on an L2 block explorer means it is final. For optimistic rollups, the transaction can still be reverted if a fraud proof is submitted within the challenge window. For ZK-rollups, the transaction is final on the L2 once the batch is proven and posted to L1, but some block explorers may show unproven transactions as confirmed. The page Once on an L2 Explorer Does Not Mean Final explains how to check actual finality status.
Risk also lives in the bridge contracts themselves. Upgrade bugs, paused contracts, and self-destructed liquidity pools have caused total losses in the past. L2Beat categorizes rollup risk and assigns stage ratings (Stage 0, Stage 1, Stage 2) based on whether the rollup has working fraud proofs, decentralized sequencing, and permissionless exit. The page How L2Beat Tracks Rollup State and Risk Stages explains what each stage actually means for a user’s safety and why no major rollup has reached Stage 2 yet.
Real errors happen too. If you have seen “transaction underpriced” on an L2 during an L1 gas spike, or “gas estimation failed” when bridging to a ZK-rollup, or an “output root missing” error on a centralized bridge frontend - those are not user mistakes. They are symptoms of how the underlying systems handle congestion. The error inventory at the top of this page lists the most common ones, and the spoke pages on specific bridges and sequencer behavior address them directly.
What the pillar page accomplishes and where to go next
This page has mapped the entire Layer 2 landscape: the two core architectures, the major chains, the bridging tradeoffs, the appchain frameworks, and the risks that cut across all of them. Every section above connects to a dedicated spoke page that answers one question completely.
If you are deciding which L2 to deploy an application on, start with Arbitrum vs Optimism Which Should I Deploy On. If you are choosing between ZK and optimistic rollups for an institutional context, read ZK Rollup vs Optimistic Rollup for Institutional Use. For developers evaluating appchain frameworks, OP Stack vs Arbitrum Orbit for Launching an Appchain covers that decision in depth. And for anyone who has ever wondered why a bridge transaction is stuck for a week, Bridge Stuck Pending Fraud Proof Window 7 Day Delay explains the mechanism and the workarounds.
Layer 2 networks are not a solved problem. They are evolving rapidly, with new proving systems, data layers, and governance models appearing every quarter. The pages in this cluster will stay updated as the tools change. The goal is to give you enough grounding to make informed decisions - not to predict which chain will win, but to understand what each one actually guarantees.
Not financial advice. mycoin.hk publishes market data and general information about digital assets. Crypto assets are volatile and you can lose everything you put in. Nothing here is a recommendation to buy, sell or hold, and we make no price predictions.
Prices are sourced from third parties and may be delayed or wrong. Verify anything you intend to act on against a primary source.