Is it difficult to scale blockchain? From Bitcoin to Ethereum

The growth of DeFi business took off in 2020, with TVL increasing 40 times from $600M to $25B. This can be intuitively felt from one phenomenon: DEX increased from thousands of transactions per day to hundreds of thousands last year, and Uniswap's daily transaction volume even exceeded that of the old centralized exchange Coinbase.

The transaction volume on Ethereum continues to rise, but the block capacity is beginning to be tight, and the mining pool chooses GasPrice bidding to sort transactions to maximize its profits. At the same time, considering the technical problems of DeFi business and the psychological problems of FOMO, everyone is willing to pay a higher price to make transactions run faster. The high mining fee makes us move further away from the original intention of inclusive finance.

So why doesn’t Ethereum choose to directly increase the block capacity to carry more transactions? The reason is simple. The core limitation of blockchain is that “every transaction must be verified and processed by every node in the network”. If the block size is increased several times, more small nodes will withdraw and gradually move towards centralization.

Is it difficult to expand the capacity of blockchain?

As early as 2010, Satoshi Nakamoto had discussed the issue of capacity expansion. Initially, Satoshi Nakamoto set the block capacity to 32M in the Bitcoin source code, but later the block transaction volume was very small, so the capacity was set to 1M. He also believed that if there was a need for capacity expansion, then it would be fine to set the block height in the code to automatically upgrade the block capacity.

However, Bitcoin expansion was not so smooth later, and the main disagreement was whether to adopt "Segregated Witness and Lightning Network" or "Increase block capacity". The former believed that "increasing block capacity requires higher bandwidth and storage, which will gradually lead to the centralization of nodes", and hoped to use the two technologies of Segregated Witness and Lightning Network to complete the expansion without increasing block capacity.

Note: This is also why Eth2 hopes to adopt PoS to lower the node threshold, so that more people can become validators and realize the ideal of a decentralized world ledger.

Of course, this was no longer a technical issue in the end, but rather rose to an ideological battle. After experiencing the inconsistency and distrust of the Hong Kong Consensus, it eventually led to a hard fork, forming the current two expansion routes of Bitcoin Core and Bitcoin Classic.

The Bitcoin expansion process is like a drama. We can see expansion technologies such as SegWit, Schnorr Signature, and Lightning Network being implemented, but there has not been any qualitative change.

• SegWit: transaction data is compressed to 25%;

• Schnorr Signature: Single signature data is compressed to 50%; Multi-signature: Multiple transactions are compressed into the size of a single transaction;

• Lightning Network: Lacks peer-to-peer small payment scenarios, and the asset size is only 1000+ BTC.

Later, Joseph Poon, the author of the Lightning Network white paper, switched to the Ethereum community and published the Plasma white paper with Vitalik in 2017, attempting to solve Ethereum's scalability problem.

Plasma advocates moving the computation and storage of transactions to Layer2, then compressing multiple transactions into a block, and submitting the Merkle root of the block to the Layer1 Plasma contract, thereby increasing throughput and reducing transaction costs.

Layer2 relies on the Layer1 Plasma contract as a security trust and arbitration layer. If there is a problem/maliciousness in Layer2, users can use the Layer1 Plasma contract to withdraw funds. To withdraw funds, you need to wait for a challenge period of about 2 weeks. During this period, anyone can submit a "fraud proof" to prove that the withdrawal is invalid and receive a reward. On the contrary, if there is no problem, wait for the challenge period to end and withdraw smoothly.

Plasma was once the hope of the community. We can see many Plasma variants such as OmiseGo, Matic, Loom, etc. However, due to some defects, it did not gain the final favor of the community:

• Due to the lack of data availability, users have to rely on Plasma trusted node data or constantly monitor the network for malicious intent, and cannot obtain security equivalent to Layer 1.

• The withdrawal cycle is long and relies on a two-week fraud proof challenge period. If someone does something malicious and no one discovers and submits proof of the malicious (fraud) behavior during the challenge period, there is a risk of losing user funds.

The above paragraph may be a little difficult to understand, let's take an example of data availability:

From the past expansion of Bitcoin and Ethereum, we have gradually found what the ideal expansion plan should look like

• Maintain the block size to avoid centralization;

• Supports data availability, users can reconstruct transaction status from L1;

With the 2019 Ethereum hard fork upgrade - Istanbul, we are getting closer to the ideal expansion plan. This upgrade includes two Layer 2-friendly EIP proposals

• EIP-1108: Make SNARKs and STARKs verification operations cheaper;

• EIP-2028: Reduce Calldata Gas consumption so that L2 data can be uploaded to the chain at low cost in the form of Calldata.

The Current State of Layer2 Rollup

What is Rollup?

Rollup has many Chinese translations, one of which is jokingly translated as "spring rolls". You can imagine the feeling of rolling up food and eating it in one bite. In fact, Rollup compresses transaction data and puts it on Layer1, so that the same block capacity can accommodate more transactions, expanding 30 TPS to 3000 TPS level, and combined with Eth2 Sharding expansion, the effect can be comparable to the current Visa level.

How to compress data?

Compression is the essence of Rollup. Clever compression techniques can accommodate more transactions. At the same time, compressing data on the chain can provide "data availability" and ensure the security of user assets.

A simple Ethereum transaction (such as sending ETH) usually consumes about 110 bytes. However, sending ETH on Rollup only consumes about 12 bytes. Let's take the receiving address and transaction signature as an example:

• To: Use "index" instead of 20-byte addresses (for example, if an address is the 4,527th address in the "state tree", we can use index 4,527 to represent it);

• Signature: Using the characteristics of zero-knowledge proof, thousands of transaction signatures are converted into zero-knowledge proofs, and the validity of thousands of transactions is confirmed by checking the validity of the proofs.

Comparison of Rollup Solutions

There are two main types of Rollup solutions: ZK Rollup and Optimistic Rollup. The former uses proof of validity, while the latter uses proof of fraud. The concepts behind these two proof methods are different:

• Proof of validity: uses mathematical methods to ensure fund security, smooth user experience and privacy, but cannot support general EVM calculations in the short term;

• Fraud proof: supports general EVM calculations (to facilitate Ethereum project migration), but zero-knowledge proof cannot be adopted in the short term, so a compromise is made to choose fraud proof to ensure the safety of funds.

Finally, this article mentions some different expansion plans in the past in order to describe the development history of Layer2, but this does not mean that the ideas of these plans are failures. In addition to technical issues, commercial feasibility is also worth pondering, as Vitalik commented on these two Rollup solutions: