Vitalik Buterin, co-founder of Ethereum, has revealed the upcoming steps for simplifying the protocol and reducing the resource load on nodes, which he refers to as “The Purge.” This stage in the Ethereum transition involves removing old and unnecessary network history to streamline the network over time.
In addition to reducing the storage of historical data, The Purge also significantly decreases the hard disk requirements for node operators and reduces the technical debt of the Ethereum protocol.
During the Dencun hard fork, Buterin introduced Ethereum Improvement Proposal (EIP)-6780, which eliminated most of the “SELFDESTRUCT” code functions. This action simplified the protocol by removing complexity and enhancing security guarantees. As a result of implementing EIP-6780, each Ethereum block will now have a higher number of storage slots. Buterin hopes that a future EIP will completely eliminate the SELFDESTRUCT code.
To limit the amount of historical data stored, The Purge will introduce history expiration through EIP-4444. This allows nodes to prune historical blocks that are over a year old. The historical data will only be necessary when a peer needs to synchronize with the head of the chain or when specifically requested.
Therefore, when new blocks are confirmed, a fully synced node will not require historical data older than 365 days. Buterin believes that EIP-4444 can greatly enhance the decentralization of Ethereum nodes.
Buterin also shared that Geth, Ethereum’s client software, has recently removed thousands of lines of code by dropping support for pre-Merge (PoW) networks. Additionally, after the Dencun hard fork, there will be an 18-day storage window for blobs, reducing the node data bandwidth to 50 gigabytes.
Furthermore, Buterin discussed the importance of purging precompile Ethereum contracts. These contracts are used to implement complex forms of cryptography that cannot be executed by the Ethereum Virtual Machine (EVM). However, the demand for these contracts has decreased over time, and they have become a “key source of consensus bugs and a significant challenge for new EVM implementations,” according to Buterin.