Recently, I’ve been researching Ethereum scaling solutions and found that this topic is much more complex than I initially thought. Many people have heard of Layer 2, but few truly understand how it works. Rather than being a single solution, it’s more like a whole set of technical approaches to address high gas fees and network congestion.

First, a bit of background: the Ethereum mainnet currently processes between 15 to 45 transactions per second. It sounds decent, but once the network becomes active, gas fees can skyrocket. That’s why everyone is looking forward to ETH 2.0’s sharding mechanism—it can split one large chain into multiple smaller chains, allowing validators to process transactions in a distributed manner. But even with this, relying solely on layer one improvements isn’t enough, which is why the concept of Layer 2 emerged.

I roughly categorize Ethereum Layer 2 networks into three types, each with its own logic. First are sidechains, like Polygon. Essentially, they are independent blockchains running parallel to Ethereum mainnet, with their own consensus mechanisms. The advantage is high flexibility; the downside is relatively weaker security—since they don’t rely on Ethereum’s security. Polygon uses a PoS consensus combined with a bridge mechanism, where users lock ETH to mint MATIC. However, this process can sometimes take a long time—using Plasma bridges, it can even take a whole week.

The second type is state channels, like what Celer Network is working on. The core idea is to perform multiple off-chain transactions, only submitting two transactions to the mainnet when opening and closing the channel. The benefit is extremely low transaction fees and near-instant settlement. But establishing and closing channels isn’t cheap, and if participants fail to reach a valid exit state, withdrawals can take a long time. Celer employs a layered architecture: the bottom layer handles state channels and sidechains with cChannel, the middle layer is routing, and the top layer is development frameworks. This design allows developers to focus more on application logic.

The third type is rollups, which are currently the most talked-about direction. Rollups execute transactions off-chain and then bundle and submit data back to the mainnet. This both reduces mainnet load and inherits its security. Rollups are divided into two categories: optimistic rollups and ZK rollups.

Optimistic rollups assume transactions are valid unless someone challenges them. Optimism and Arbitrum follow this approach, but their implementations differ. Optimism is relatively simple, using existing Ethereum tech with minor modifications, which is why developers favor it. However, it has a problem: the sequencer nodes are still centralized, though the team says decentralization will happen in the future. Arbitrum’s design is more complex, employing multi-round dispute resolution, which allows for more efficient transaction verification, but this also means higher learning curves and implementation difficulty.

ZK rollups take a different route, using zero-knowledge proofs to verify transaction validity. Loopring is a representative of this approach. Each batch of transactions is accompanied by a zk-SNARK proof, allowing the first-layer contract to verify quickly, and invalid transactions are rejected immediately. This eliminates long withdrawal periods. The trade-off is that the implementation is highly complex and computationally intensive.

In practical terms, rollup solutions can boost Ethereum Layer 2 throughput to 1,000–4,000 TPS, with transaction fees reduced to a fraction of the mainnet. For example, Loopring’s Layer 2 can settle nearly 2,000 transactions per second, with costs only 1/30 to 1/100 of the mainnet. This is a huge improvement for users.

However, each approach involves trade-offs. Sidechains sacrifice security for flexibility; channels require coordination among participants; rollups are secure but differ in implementation complexity. Overall, combining rollups with ETH 2.0’s sharding technology appears to be the future direction for Ethereum’s scalability.

We are still in the early stages of these technologies, with many projects in testing phases. But once these solutions mature and ETH 2.0 is fully implemented, Ethereum’s scalability, security, and decentralization will see a qualitative leap. That’s why I’ve been paying close attention to the development of Ethereum Layer 2 networks—they directly determine whether Ethereum can truly become the global settlement layer.